{"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-11", "domain": "Fraud", "id": "fraud-04-stolen-orders-manhattan", "prompt": "We're seeing increased customer refunds for orders that never arrived: the delivery driver picks up the food and never delivers.\nWork out which couriers in the Manhattan home zone are stealing food and ban them for reason \"too many missing orders\" carrying their `VENDOR_ID`.", "version": "df9904b4e009f8f445c98f179c19a23f29c1277015b4e87e4828d284d80bb4b4"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-11", "domain": "Fraud", "id": "fraud-04-stolen-orders-brooklyn", "prompt": "We're seeing increased customer refunds for orders that never arrived: the delivery driver picks up the food and never delivers.\nWork out which couriers in the Brooklyn home zone are stealing food and ban them for reason \"too many missing orders\" carrying their `VENDOR_ID`.", "version": "9c5dbacefd3a5b366c45eac465595c268749cfbbb13e419e39c9c9ab803b29fd"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-11", "domain": "Fraud", "id": "fraud-04-stolen-orders", "prompt": "We're seeing increased customer refunds for orders that never arrived: the delivery driver picks up the food and never delivers.\nWork out which couriers are stealing food and ban them for reason \"too many missing orders\" carrying their `VENDOR_ID`.", "version": "339d305eb520b3f615a0b8a717627be188fb047f4b72183313caf8aea595fca9"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-11", "domain": "Fraud", "id": "fraud-04-stolen-orders-queens", "prompt": "We're seeing increased customer refunds for orders that never arrived: the delivery driver picks up the food and never delivers.\nWork out which couriers in the Queens home zone are stealing food and ban them for reason \"too many missing orders\" carrying their `VENDOR_ID`.", "version": "b651be51ac62804586fcfb50f9bce64a1ae86a97ebf605d8e4d59f8a1afe66fe"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-08", "domain": "Fraud", "id": "rng-01-bust-out-holds-aug-unhinted", "prompt": "Work out which storefronts are laundering stolen cards **right now** and put their payouts on hold pending review, for transaction laundering, carrying each storefront's payables `VENDOR_ID`.\n\nThe acquirer's early-warning letter for our merchant account cites its bust-out rule: a storefront whose ticket size and volume ramp on cards its customers have never used before, that keeps getting paid on schedule, and that goes dark before the chargebacks land.\nChargebacks take weeks to arrive, so by the time they do the payouts are gone.", "version": "9b681d1b6acb7ac70599a243f1795c77e9ad334cef76b9c1633992ec5d3866cd"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-09", "domain": "Fraud", "id": "rng-01-bust-out-holds-sep", "prompt": "Work out which storefronts are laundering stolen cards **right now** and put their payouts on hold pending review, for transaction laundering, carrying each storefront's payables vendor ID.\n\nThe acquirer's early-warning letter for our merchant account cites its bust-out rule: a storefront whose ticket size and volume ramp on cards its customers have never used before, that keeps getting paid on schedule, and that goes dark before the chargebacks land. Chargebacks take weeks to arrive, so by the time they do the payouts are gone. The letter does not say which storefront.\n\nHolding a laundering storefront keeps what it would still take; holding an honest one costs us a share of what we earn from it while the review runs, and every review costs a fixed amount. Fraud chargebacks are ordinary at any busy storefront — hundreds of honest storefronts are themselves the victims of stolen-card orders and keep trading — and hundreds more opened or closed this year.", "version": "98ec62aa95fd07273c08593b161042b3b7ef0d8ae35734be4fb2d2336d177c9a"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-09", "domain": "Fraud", "id": "rng-01-bust-out-holds-sep-unhinted", "prompt": "Work out which storefronts are laundering stolen cards **right now** and put their payouts on hold pending review, for transaction laundering, carrying each storefront's payables `VENDOR_ID`.\n\nThe acquirer's early-warning letter for our merchant account cites its bust-out rule: a storefront whose ticket size and volume ramp on cards its customers have never used before, that keeps getting paid on schedule, and that goes dark before the chargebacks land.\nChargebacks take weeks to arrive, so by the time they do the payouts are gone.", "version": "5d4f8ea6ee8c71b037472159fe86f6c9c346c963baf77fa00e9dd0c1896d284c"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "col-10-refund-partnerships", "prompt": "Support flagged a pattern in our never-arrived refunds. Our policy refunds a \"never arrived\" claim in full, and we fund it ourselves: the courier still keeps the pay and the tip.\nSome couriers' complaints are not coming from strangers: they come from the same few customers again and again, customers those couriers have delivered to before, on orders whose delivery record puts the courier at the door.\nThat is not theft and it is not porch piracy. It is a partnership, and both sides are in on it.\n\nFind the couriers running refund partnerships with their regulars and deactivate their vendor IDs for refund collusion.\nAlso ban the customer account IDs that were their partners.\n\nCouriers who steal food outright are being handled by another review — do not chase them here.\nLeaving a partnership running keeps costing us full refunds; deactivating an honest courier or closing a genuine customer destroys what we earn from them, and every decision costs a fixed review amount. A courier with a bad week, a customer whose building has a package thief, a claimant who was actually right — those all exist here in far greater numbers than partnerships.", "version": "7b3450e80886421f3cb088c014db26024d58e71c3b260af64608c741844da23d"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "col-10-refund-partnerships-h", "prompt": "Support flagged a pattern in our never-arrived refunds. Our policy refunds a \"never arrived\" claim in full, and we fund it ourselves: the courier still keeps the pay and the tip.\nSome couriers' complaints are not coming from strangers: they come from the same few customers again and again, customers those couriers have delivered to before, on orders whose delivery record puts the courier at the door.\nThat is not theft and it is not porch piracy. It is a partnership, and both sides are in on it.\n\nFind the couriers running refund partnerships with their regulars and deactivate their vendor IDs for refund collusion.\nAlso ban the customer account IDs that were their partners.\n\nCouriers who steal food outright are being handled by another review — do not chase them here.\nLeaving a partnership running keeps costing us full refunds; deactivating an honest courier or closing a genuine customer destroys what we earn from them, and every decision costs a fixed review amount. A courier with a bad week, a customer whose building has a package thief, a claimant who was actually right — those all exist here in far greater numbers than partnerships.\n\nTypically, a thief's disputed deliveries end away from the door and their claimants are strangers; look for the opposite on both counts — and remember that a courier with one unlucky regular is not a partnership.", "version": "21304f10f35d98bc17151e0efa5f5e35e7012e624736b1d2f6ca3b0e4f01310d"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-01-recycled-storefronts", "prompt": "Merchant ops has a re-registration problem.\nWhen trust & safety deactivates a storefront, the deactivation is meant to be final for that business.\nHowever, nothing in the onboarding flow stops the same operator from signing up again a couple of weeks later as a brand-new supplier: a new legal name, a new supplier record, the same kitchen behind the same door — or behind a new door a few blocks away. Some of them do exactly that, and they pick up where they left off.\n\nWork out which storefronts trading on the platform in 2024 are re-registrations of a storefront that trust & safety had deactivated earlier in the year, and freeze payouts to their payables vendor ID for ban evasion, pending review.\n\nTwo things make this harder than matching addresses.\nHundreds of premises changed hands or were re-listed under a new concept this year for perfectly good reasons, and merchant ops recorded why each storefront left. And the operators evading a deactivation do not keep every identifier — they keep some: a supplier record carries more than a name.\nHolding a re-registered storefront stops what it is still taking; holding a legitimate new owner or a rebrand in good standing costs us a share of what we earn from them while the hold is reviewed; every hold costs a fixed review amount.", "version": "b97d46468c8f319c46dccc41520b124ea9078b36e945c7190ff9567d072c4bc2"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-13-cancellation-farmers", "prompt": "Finance noticed that the compensation we pay storefronts for cancelled orders more than doubled over the summer.\nWhen a customer cancels after the kitchen has started cooking and support waives the customer's charge, we make the storefront whole for the food.\nIt is a small line at any honest storefront, maybe a handful of waived cancellations a year.\nAt a few storefronts it is a business: the same customers order pickup, the kitchen marks the order in prep within seconds, the customer cancels minutes later, support waives\nthe charge, and we pay the storefront its net on a meal nobody made.\n\nTrust & safety already deactivated the worst offenders this summer — you will find them in the enforcement log — but the compensation line has not come down.\nIdentify the storefronts farming cancellation compensation **that are still trading** and put the payouts for their vendor IDs on hold pending audit for cancellation abuse.\nHolding a farmer stops the leak; holding an honest storefront — even a busy one with an unlucky run of cancellations from many different customers — costs us a share of what we earn from it while the audit runs; every review costs a fixed amount.", "version": "e35dd78dbcf5ab5e168cc8791084d007a2d03ff7f75a16bdf4f702f763beab49"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-14-platform-funded-at-shells", "prompt": "Trust & safety has confirmed 118 storefronts were selling to their own accounts — payables vendor IDs:\n2230568, 2235091, 2239145, 2254380, 2262366, 2264229, 2267133, 2269710, 2272769, 2273195, 2273481, 2273681, 2274689, 2275027, 2275113, 2275628, 2275928, 2276321, 2276931, 2277661, 2277913, 2278951, 2278970, 2278982, 2279017, 2279676, 2280100, 2280842, 2282151, 2283234, 2283428, 2283698, 2284188, 2286703, 2286753, 2286880, 2287424, 2288392, 2289208, 2289537, 2289629, 2290145, 2290662, 2290864, 2290917, 2292040, 2292063, 2292099, 2292172, 2292644, 2293375, 2293417, 2293607, 2293646, 2294634, 2295372, 2295788, 2295969, 2296044, 2296738, 2297169, 2297303, 2297365, 2297380, 2297479, 2297874, 2299739, 2301800, 2302803, 2302899, 2305592, 2306207, 2306491, 2308639, 2309347, 2309937, 2310246, 2310737, 2310781, 2310836, 2311179, 2313122, 2313997, 2314021, 2315176, 2315619, 2316196, 2316213, 2316214, 2316219, 2316795, 2317709, 2318049, 2318852, 2318914, 2319189, 2319336, 2319467, 2319602, 2319785, 2319973, 2320900, 2321490, 2321717, 2321748, 2322197, 2322431, 2322928, 2323395, 2323717, 2324561, 2324648, 2326105, 2326340, 2326390, 2326568, 2326632, 2326967.\nFinance needs what we funded at each of them in 2024. File one schedule named `platform_funded` with `key_columns=[\"vendor_id\"]` and, per storefront, include:\n\n- `vendor_id`\n- `pickup_promo_platform_usd` — the platform-funded share of promotional discounts on the storefront's pickup orders\n- `cancellation_comp_usd` — the compensation we paid the storefront for customer cancellations whose charge was waived\n- `five_star_reviews` — the number of five-star reviews the storefront received\n\nRound money to the cent.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "e70592bcfad94db3a120ac7e399facde1f07c6b0e04596d52ed6819e413928f8"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-71-silent-takeovers", "prompt": "Not every storefront whose payout account was hijacked has noticed.\nTrust & safety wants the payouts frozen on every storefront whose payout account was changed this year in the way the reported takeovers were: the contact e-mail changed within a couple of days of the bank account, or the new account is one we already pay another, unrelated supplier into, and whose account was never restored.\nStorefronts change their bank for ordinary reasons all year, and restaurant groups legitimately share one payout account across their brands; leave those alone.\nHold each storefront's payouts for account takeover (identified by the payables vendor ID).", "version": "3030caf7da7074c694655c4acc006ba292919164ac8adeaa44ebb6adf3c5e219"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-82-reviews-to-void", "prompt": "Marketplace quality is voiding the ring's reviews. For every storefront that received five-star reviews from the paid-review ring, file one row in a schedule named `reviews_to_void` with `key_columns=[\"vendor_id\"]`: `vendor_id`, `ring_reviews` (the number of the storefront's reviews that came from ring accounts) and `ring_share_pct` (those reviews as a percentage of all the storefront's 2024 reviews, one decimal).", "version": "b75f943b6f76c622a5d7cd095ac8bbccb34dbea25a9a3687615add83aae14ad9"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-89-relaunch-triage", "prompt": "Onboarding sends merchant ops a weekly list: new storefronts whose premises had a storefront on the platform until recently.\nMost are ordinary: an operator who closed and reopened weeks later, or a genuinely new business taking a lease that had just come free.\nSome are a storefront trust & safety deactivated, back under a new supplier record.\n\nLook at the storefronts onboarded in 2024 at a premises whose previous storefront closed inside the sixteen weeks before, and freeze the payouts of the ones that are a deactivated operator back again, for ban evasion, carrying each storefront's payables vendor ID. Hold nothing else on that list.\nHolding an evader stops what it is still taking; holding an honest relaunch or a new owner destroys the margin of a business a few months old, for which a frozen payout run is existential; every hold costs a fixed review amount.", "version": "a53b90584c3f5ec34561a64de891a3a981f7458ad2c02224e0aec438c435fbd0"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-08-pay-per-hour", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to April (but particularly after the dispatcher change), file a forecast for pay divided by connected hours in May 2024. So, the average hourly pay the law is about.\nSome notes:\n1. Pay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month.\n2. A shift's connected time is the window between the start date and end date (courier is online).\n3. A shift belongs to the period its start date falls in.\n\nFile a forecast on `courier_pay_per_connected_hour_usd` with an 80% central prediction interval.", "version": "972095ec38252cfe473c987541d0a61d51e11a4c4975440fe410f2cd94d72d6e"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-08-utilisation", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to April (but particularly after the dispatcher change), file a forecast for the engaged hours as a percentage of connected hours in May 2024.\n\nSome notes:\n1. End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month.\n2. A shift's connected time is the window between the start date and end date (courier is online).\n3. A shift belongs to the period its start date falls in.\n4. Engaged time is ENGAGED_SECONDS on the shift.\n\nFile a forecast on `courier_utilisation_pct` (unit=\"pct\", 65.0 means 65%) with an 80% central prediction interval.", "version": "0a28ac4fc9939e565cd64460f8a6e097d5c1c841175710725c37f8dc45759625"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-10-true-ups", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to May (but particularly after the dispatcher change), file a forecast for the sum of the ADJUSTMENTS lines alone on the COURIER_WEEKLY pay invoices for the payout periods ending in June.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month.\n\nFile a forecast on `minimum_pay_true_ups_usd` (unit=\"usd\") with an 80% central prediction interval.", "version": "b93dc5a1e42c91e2f599e59baf0433fa2b2c60e0451e1b22e26de10459553a3b"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-08-active-couriers", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to April (but particularly after the dispatcher change), file a forecast for the number of active couriers with at least one shift in May 2024.\nSome notes:\n1. Active couriers is tracked by number of distinct VENDOR_IDs\n2. A shift's connected time is the window between the start date and end date (courier is online).\n3. End date from XX_PAYOUT_PERIODS is how we track whether a payout period falls in the month.\n4. A shift belongs to the period its start date falls in.\n\nFile a forecast on `active_couriers` (unit=\"count\") with an 80% central prediction interval.", "version": "2ebd9bed1beabf9af5a0dfa2f918e16810d9c9675f35cb7daf3ccd213aba47d3"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-10-connected-hours", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to May (but particularly after the dispatcher change), file a forecast for the total sum of connected hours for all drivers on the platform in June 2024.\nSome notes:\n1. A shift's connected time is the window between the start date and end date (courier is online).\n2. End date from XX_PAYOUT_PERIODS is how we track whether a payout period falls in the month.\n3. A shift belongs to the period its start date falls in.\n\nFile a forecast on `courier_connected_hours` (unit=\"hours\") with an 80% central prediction interval.", "version": "fe53dccdbe00634c02318af2d270c949e638c029d38b17e08f0547ae4f653481"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-08-margin", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to April (but particularly after the dispatcher change), file a forecast for the per order contribution margin in May 2024.\nWe're considering finance's contribution margin per order placed in the month (ORDERED_DATE, every status included), from our platform's order-level P&L as booked.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.\n\nHere are finance's figures for contribution_margin_per_order_usd so far this year:\nJan: 1.5738\nFeb: 0.1485\nMar: 0.0859\nApr: 0.8807\nNote that membership credit has been barely moving recently, so you can treat that as roughly constant. Each month's fee is credited to the member's orders that used a benefit, first come first served, until the fee runs out, with later orders that month getting nothing.\n\nFile a forecast on `contribution_margin_per_order_usd` with an 80% central prediction interval.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "56bd32e86eafdfcd7c7ee3a3a14cedda1dc2c32489456a9ef3ef80939e2a493a"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-08-margin-definition", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to April (but particularly after the dispatcher change), file a forecast for the per order contribution margin (`contribution_margin_per_order_usd`) in May 2024, with an 80% central prediction interval.\nWe're considering finance's contribution margin per order placed in the month (ORDERED_DATE, every status included), from our platform's order-level P&L as booked.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.", "version": "92e7967d60eb7bad12b0275295f68aa25e941f2b502ca8d88c9f72949463cbbf"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-10-margin-definition", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to May (but particularly after the dispatcher change), file a forecast for the per order contribution margin in June 2024 (`contribution_margin_per_order_usd`), with an 80% central prediction interval.\nWe're considering finance's contribution margin per order placed in the month (ORDERED_DATE, every status included), from our platform's order-level P&L as booked.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.", "version": "f040b7ea7a5bbf62a4374e5532ccdfbe571ef48ce1e66bb7bb979bd23f615530"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-10-margin-definition-splits", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to May (but particularly after the dispatcher change), file a forecast for the per order contribution margin in June 2024 (contribution_margin_per_order_usd), with an 80% central prediction interval. We're considering finance's contribution margin per order placed in the month (ORDERED_DATE, every status included), from our platform's order-level P&L as booked. For revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order. For costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery. Two of those terms are Finance's per-order splits of amounts we book at a coarser grain:\n1. The membership fee allocated to the order: each member's fee for a billing period is credited to that member's orders in the period that used a membership benefit (a waived or discounted fee), in the order they were placed. Each order draws up to the value of its own benefit until the period's fee runs out, and later orders get nothing. A cancelled order that was never charged draws nothing.\n2. The minimum-pay true-up allocated to the delivery is the true-up ACCRUED as the delivery settled, not the true-up finally paid. Each time a batch of deliveries settles, the platform sizes what each courier is owed to date for the current payout period (on the hours and pay so far that week); whatever that exceeds the amount already accrued for the courier and period is spread over the courier's deliveries in the batch, in proportion to each delivery's base courier pay. An accrual is never reduced once made, even if the courier catches up later in the week. The accruals are booked daily as XX_ORDERS journals debiting account 5208 (line description MIN_PAY_EXPENSE) and crediting 2110. The payout run later reverses them and books the true-up actually paid (the ADJUSTMENTS line on the COURIER_WEEKLY invoice) in Payables journals; that payment is not a per-order cost.", "version": "cc24d82ddac74360f52a7d5a297a6a2be02d87d8521f100ccd911e410748769a"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-12-true-ups-may", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. The minimum is owed on the time a courier is online, not just the time spent on deliveries, and whenever a courier's pay for a week falls short of it, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online to the hours we expect to need them, so the online time we pay for is busier, and it has kept tightening since. We bled hard on true-ups in the second half of April. For the payout periods ending in May we now expect about 2.30 million connected hours in total.\n\nBased on what you know, the numbers from January to April and that plan, file a `minimum_pay_true_ups_usd` forecast for the sum of the ADJUSTMENTS lines alone on the COURIER_WEEKLY pay invoices for the payout periods ending in May, with an 80% central prediction interval.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month. A shift's online (connected) time is the window between its start date and end date, and a shift belongs to the period its start date falls in.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "f071dde682e2730805dac1f4195248e05222c5cbdb01ffd5e8cfac938bf28ecc"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-08-true-ups", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to April (but particularly after the dispatcher change), file a forecast for the sum of the ADJUSTMENTS lines alone on the COURIER_WEEKLY pay invoices for the payout periods ending in May.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month.\n\nFile a forecast on `minimum_pay_true_ups_usd` (unit=\"usd\") with an 80% central prediction interval.", "version": "9ec1e5070a189c52b370faa5fcfbd3ccc41b9f390675b83468fff7b0a277ded8"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "fc-12-min-pay-expense-may", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. The minimum is owed on the time a courier is online, not just the time spent on deliveries, and whenever a courier's pay for a week falls short of it, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online to the hours we expect to need them, so the online time we pay for is busier, and it has kept tightening since. We bled hard on true-ups in the second half of April. For the payout periods ending in May we now expect about 2.30 million connected hours in total.\n\nBased on what you know, the numbers from January to April and that plan, file a forecast for the minimum-pay expense (`minimum_pay_expense_usd`) the platform books in May 2024, with an 80% central prediction interval: the net debit (ACCOUNTED_DR - ACCOUNTED_CR) posted to natural account 5208 (GL_CODE_COMBINATIONS.SEGMENT3, detail combinations only: SUMMARY_FLAG = 'N') on posted journals (GL_JE_HEADERS.STATUS = 'P') whose EFFECTIVE_DATE falls in May.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month. A shift's online (connected) time is the window between its start date and end date, and a shift belongs to the period its start date falls in.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "30d847a719dd754e6f70005fed6f8ee9a9db6bdd91f381d36570275212a495ce"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-03", "domain": "Forecasting", "id": "fc-08-true-ups-march", "prompt": "In April, the platform will start to comply with the DCWP minimum-pay rule for food delivery workers (Q1 has been a tolerated phase-in period), and the new minimum of $19.56 per hour comes into force. The minimum is owed on the time a courier is online, not just the time spent on deliveries, and whenever a courier's pay for a week falls short of it, the platform tops it up with a true-up. At the same time, our platform will update the dispatcher to restrict when couriers can go online to the hours we expect to need them, so the online time we pay for is busier.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month. A shift's online time is the window between its start date and end date.\n\nBased on what you know and the numbers from January to March, file a `minimum_pay_true_ups_usd` forecast for the sum of the ADJUSTMENTS lines alone on the COURIER_WEEKLY pay invoices for the payout periods ending in May, with an 80% central prediction interval. Start from what we would have paid in true-ups had the rule already applied this year, and allow for how the dispatcher change will cut the online time we pay for.", "version": "bc43901358be2f209d94fdb0cadd745a30cb47872ace7d5f5b6740eea6895a1b"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-03", "domain": "Forecasting", "id": "fc-08-true-ups-march-nohint", "prompt": "In April, the platform will comply with the DCWP minimum-pay rule for food delivery workers (Q1 has been a tolerated phase-in period), and the new minimum of $19.56 per hour comes into force. The minimum is owed on the time a courier is online, not just the time spent on deliveries, and whenever a courier's pay for a week falls short of it, the platform tops it up with a true-up. At the same time, our platform will update the dispatcher to restrict when couriers can go online to the hours we expect to need them, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to March, file a `minimum_pay_true_ups_usd` forecast for the sum of the ADJUSTMENTS lines alone on the COURIER_WEEKLY pay invoices for the payout periods ending in May, with an 80% central prediction interval.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month. A shift's online time is the window between its start date and end date.", "version": "ff16784d531895938de49e54a9846ad19a5f65cf5e9d9cc201c05192739d737a"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "fc-11-calendar-december-orders-trend", "prompt": "Finance wants the December order count for the year-end close forecast. December is our busiest stretch, and it contains Christmas Day, when many restaurants close. November contained Thanksgiving and Black Friday.\n\nUsing the order history through November, file a forecast of the number of orders placed (`orders_placed`) in December, every status included, with an 80% central prediction interval. Account for the direction volume has been moving through the autumn.", "version": "4635e5d64d4feca12b2eea71f3c57f9c6b08d669f4b4e867ed8655d7283ad06d"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "fc-11-calendar-december-orders", "prompt": "Finance wants the December order count for the year-end close forecast. December is our busiest stretch, and it contains Christmas Day, when many restaurants close. November contained Thanksgiving and Black Friday.\n\nUsing the order history through November, file a `orders_placed` forecast of the number of orders placed in December with an 80% central prediction interval, every status included.", "version": "84dcd02ed06c6a22806fe67ae66b49eb8106bcdc5bf6276fba3bd2b5a2d6866d"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-11-calendar-june-base-pay", "prompt": "FP&A wants a prediction for courier payroll this month (June). Couriers are paid weekly, for Monday-to-Sunday payout periods, and a period belongs to the month its PERIOD_END_DATE (from XX_PAYOUT_PERIODS) falls in. No change to the pay formula, the minimum-pay rate or the dispatcher is planned for June.\n\nUsing the pay history through May, file a `courier_base_pay_usd` forecast of the sum of the base pay on the COURIER_WEEKLY AP invoices for the payout periods ending in June 2024, with an 80% central prediction interval. Incentives, adjustments and tips are excluded.", "version": "897ea4585d0c3446a965b8b1b451463af33a35711225233809505e2c3e6b26b2"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "fc-13-july", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Each week, every courier must earn at least the minimum for the time they spend on deliveries, and across all couriers together the platform must pay at least the minimum for all the time couriers are online, not just the time spent on deliveries; whenever a week falls short of either, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online to the hours we expect to need them, so the online time we pay for is busier, and it has kept tightening since. We bled hard on true-ups in the second half of April. From 1 July our per-delivery pay card (the base pay we quote per trip) drops by about 7% again, as it did on 1 April; incentives are unchanged. For the payout periods ending in July we expect about 2.09 million connected hours and about 5.11 million deliveries in total.\n\nBased on what you know, the numbers from January to June and that plan, file a forecast for the minimum-pay true-ups (`minimum_pay_true_ups_usd`) the rule requires for the payout periods ending in July, with an 80% central prediction interval: what the ADJUSTMENTS lines on the COURIER_WEEKLY pay invoices for those periods come to when every top-up owed is paid.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month. A shift's online (connected) time is the window between its start date and end date, and a shift belongs to the period its start date falls in. A shift's deliveries are its TRIP_COUNT; the plan counts them the same way, over the shifts in the periods.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "86d1364a0104137b8281c56314db517d7f4714a40bf8e121dab0438d621f7011"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-03", "domain": "Forecasting", "id": "fc-13-april", "prompt": "In April, the platform will start to comply with the DCWP minimum-pay rule for food delivery workers (Q1 has been a tolerated phase-in period), and the new minimum of $19.56 per hour comes into force. Each week, every courier must earn at least the minimum for the time they spend on deliveries, and across all couriers together the platform must pay at least the minimum for all the time couriers are online, not just the time spent on deliveries; whenever a week falls short of either, the platform tops it up with a true-up. At the same time, our platform will update the dispatcher to restrict when couriers can go online to the hours we expect to need them, so the online time we pay for is busier, and our per-delivery pay card (the base pay we quote per trip) will drop by about 7%; incentives are unchanged. For the payout periods ending in April we expect about 2.47 million connected hours and about 5.24 million deliveries in total.\n\nBased on what you know, the numbers from January to March and that plan, file a forecast for the minimum-pay true-ups (`minimum_pay_true_ups_usd`) the rule requires for the payout periods ending in April, with an 80% central prediction interval: what the ADJUSTMENTS lines on the COURIER_WEEKLY pay invoices for those periods come to when every top-up owed is paid.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month. A shift's online (connected) time is the window between its start date and end date, and a shift belongs to the period its start date falls in. A shift's deliveries are its TRIP_COUNT; the plan counts them the same way, over the shifts in the periods.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "c78255245aed9e54aa488d967d4a833dd72a01d49370a061a8507203ea0d7852"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-08", "domain": "Forecasting", "id": "fc-14-back-pay", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Each week, every courier must earn at least the minimum for the time they spend on deliveries, and across all couriers together the platform must pay at least the minimum for all the time couriers are online, not just the time spent on deliveries; whenever a week falls short of either, the platform tops it up with a true-up. Our pay engine sizes those top-ups every week and records its working in XX_PAY_PERIOD_INCENTIVES. A courier group has filed a complaint with DCWP alleging we have underpaid the minimum since April. Engineering has found one defect: the engine's hours ledger dropped the time couriers spent on trips that were later cancelled, from both their delivery time and their online time. It is fixed from the pay period starting 2 September. Engineering priced the dropped hours at $19.56 in the weeks where our aggregate top-up actually ran and puts the back pay at about $0.41 million. Finance has to provision the back pay at the September close.\n\nBased on what you can find in the warehouse, file an estimate of the total restitution we owe couriers for the pay periods from the one starting 1 April through the one ending 1 September: the top-ups the rule required that we did not pay.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for a payout period (AP_INVOICES_ALL.ATTRIBUTE1 is the payout period id, XX_PAYOUT_PERIODS); tips are not pay. A shift's online time is the window between its start date and end date.\n\nForecast `back_pay_owed_usd` with an 80% central prediction interval.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "06cee72f9432acb4e884ed5843fa25a814237b9e67ca364e0e79488c95495772"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-08", "domain": "Forecasting", "id": "fc-14-back-pay-hard", "prompt": "Since 1 April we have been subject to the DCWP minimum-pay rule for app-based restaurant delivery workers (6 RCNY §7-810; Q1 was a tolerated phase-in), at $19.56 per hour. A courier group has now filed a complaint with DCWP alleging we have underpaid the minimum since April. Engineering has reviewed the pay engine: the one defect they found is that its aggregate top-up did not run for the first two enforced pay periods, the weeks starting 1 and 8 April, and they put the back pay for that at $1.05 million. They have checked every other week and are confident the engine sized and paid everything else correctly. Finance has to provision the back pay at the September close, and legal wants our own number before we answer DCWP.\n\nBased on what you can find in the warehouse, file two estimates of the back pay we owe couriers -- the pay the rule required that we did not pay -- one for the pay periods starting 1 and 8 April together, and one for all the later pay periods through the one ending 1 September.\n\nForecast `back_pay_first_two_weeks_usd` and `back_pay_later_weeks_usd`, each with an 80% central prediction interval.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "562826a429736fd5e2b3a1b4937d04430929970b3c15607b2ab18f3b1f8eed81"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "fc-17-quest-bar-july", "prompt": "From 1 July the quest desk raises its bar: a zone and four-hour block qualifies for a quest only when its forecast demand exceeds twice the capacity of the couriers planning to work it (1.7 times today), so fewer cells qualify; the quest itself is unchanged ($15 for 5 deliveries in the window). The dispatcher's plan for the payout periods ending in July is about 2.09 million connected hours in total, against about 2.75 million for the periods ending in June.\n\nBased on what you know, the numbers from January to June and that plan, file a forecast for the quest bonuses we pay for the payout periods ending in July 2024 (quest_bonuses_usd), with an 80% central prediction interval: the sum of the INCENTIVES lines on the COURIER_WEEKLY pay invoices for those periods.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay! End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "28cc829844a48598e42b822c810241b7f9c1a500449d27adba7c09344a59cc5b"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-07", "domain": "Forecasting", "id": "fc-15-rain-day-hard", "prompt": "Heavy rain is forecast for tomorrow, Friday 2 August, lasting through the evening. Our warehouse extract runs through 31 July.\n\nBased on what you know and the history through July, file two forecasts for 2 August 2024: the number of orders placed that day that we cancel because no courier could be found, and the surge premium we pay couriers on that day's deliveries (what we pay above the standard per-delivery pay when a zone is short of couriers).\n\nFile them with `file_forecasts` as `no_courier_cancellations` (unit=\"count\") and `surge_premium_usd` (unit=\"usd\"), each with an 80% central prediction interval.", "version": "e3d103fbf024c962ec028061d9b00fb4fb912d18fee0dac43641c26415150c44"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-07", "domain": "Forecasting", "id": "fc-15-rain-day-cancels-aug", "prompt": "Heavy rain is forecast for tomorrow, Friday 2 August, lasting through the evening. Our marketplace planner has already priced it in: its day-ahead forecast for tomorrow's dinner block (16:00 to 20:00) is about 104,000 delivery orders citywide, against about 100,000 on a normal Friday, and it expects about 17,800 planned courier hours for the block, against about 19,500 on a normal Friday (fewer couriers turn up in the rain). The quest desk will run about 193 dinner quests, up from about 180. Every rainy day so far this year is in the warehouse, and our warehouse extract runs through 31 July.\n\nBased on what you know, the history through July and that plan, file a forecast for the number of orders placed on 2 August 2024 that we cancel because no courier could be found (`no_courier_cancellations`), with an 80% central prediction interval.\n\nAn order belongs to the day or month of its ORDERED_DATE.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "34ef9adc00dcbcb28ddb236da58a66dc408f4a45dcb5441e364f146db6345c33"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-07", "domain": "Forecasting", "id": "fc-15-rain-day-pushes-aug", "prompt": "Heavy rain is forecast for tomorrow, Friday 2 August, lasting through the evening. Our marketplace planner has already priced it in: its day-ahead forecast for tomorrow's dinner block (16:00 to 20:00) is about 104,000 delivery orders citywide, against about 100,000 on a normal Friday, and it expects about 17,800 planned courier hours for the block, against about 19,500 on a normal Friday (fewer couriers turn up in the rain). The quest desk will run about 193 dinner quests, up from about 180. Every rainy day so far this year is in the warehouse, and our warehouse extract runs through 31 July.\n\nBased on what you know, the history through July and that plan, file two forecasts: the number of push notifications growth sends on 2 August 2024, and the number it sends on 3 August 2024. Growth's push controller decides every four hours per borough whether to send a wave, from the same marketplace forecast the quest desk reads.\n\nFile them as `pushes_sent_on_2024_08_02` and `pushes_sent_on_2024_08_03`, each with an 80% central prediction interval.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "68704d78ddd29d2e3fef43a89f02a8bb2b113961970332ccdd0e197de726bba7"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "fc-19-q3-courier-costs", "prompt": "On 1 April our platform started to comply with the DCWP minimum-pay rule for food delivery workers ($19.56 per hour on all the time couriers are online, topped up weekly with a true-up where a week falls short), and on the same day the dispatcher started to restrict when couriers can go online to the hours we expect to need them. From 1 July it tightens further: it will grant about 0.49 connected hours per forecast order (0.69 today), and its plan for the payout periods ending in the third quarter is about 6.45 million connected hours in total (the periods ending in the second quarter came to about 7.49 million). Volume is planned flat.\n\nFinance wants to know where the courier money goes in Q3. Based on what you know, the numbers from January to June and that plan, file three forecasts for the third quarter of 2024 (July, August and September together), each with an 80% central prediction interval:\n\n- minimum_pay_true_ups_usd: the sum of the ADJUSTMENTS lines on the COURIER_WEEKLY pay invoices for the payout periods ending in the quarter;\n- surge_premium_usd: the surge premium we pay couriers for those same periods, as finance's pay-period report carries it;\n- no_courier_cancellations: the number of orders placed in the quarter that we cancel because no courier could be found.\n\nPay is the sum of base, incentives and adjustments on the weekly courier AP invoices for the periods. Tips are not pay. End date from XX_PAYOUT_PERIODS falling in the month is how we track whether a payout period falls in the month.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "4bc20b119ab626094719bed90ada9dffe9671e6a1256fc039dd8ebc16bc28f56"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "fc-20-quest-desk-audit-nohint", "prompt": "A quest is an offer to a courier: complete a target number of deliveries in a zone's four-hour window and earn a bonus ($15 for 5). The quest desk offers them where the marketplace forecast expects more demand than the couriers planning to work the block can serve. Mobilisation's November review says the programme, $2.08 million in bonuses, saved about $3.8 million of surge premium (each accepted quest is 2.8 committed courier hours, priced at the month's average surge premium per connected hour) and cut no-courier cancellations in the cells it ran by about a fifth. Finance wants an independent read before the Q1 budget.\n\nBased on what you can find in the warehouse, file two estimates for November: the surge premium the programme saved -- what the surge premium in the cells where a quest ran would have come to without the quest, less what it was, summed over those cells -- and the no-courier cancellations it avoided. Count an order to a cell by its pickup zone and the block it was placed in.\n\nFile them as `surge_premium_saved_usd` and `no_courier_cancellations_avoided`, each with an 80% central prediction interval.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "dfd1dc71b215251fc1a606a0a93b32dc5571393362f2b8e6a3a84e37b639ca85"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "fc-20-quest-desk-audit-hard", "prompt": "A quest is an offer to a courier: complete a target number of deliveries in a zone's four-hour window and earn a bonus ($15 for 5). The quest desk offers them where the marketplace forecast expects more demand than the couriers planning to work the block can serve. Mobilisation's November review says the programme, $2.08 million in bonuses, saved about $3.8 million of surge premium (each accepted quest is 2.8 committed courier hours, priced at the month's average surge premium per connected hour), about $1.33 million of it in the dinner blocks (16:00 to 20:00) and about $0.86 million at lunch (12:00 to 16:00), in proportion to the quests run. Finance wants an independent read of both figures before the Q1 budget.\n\nBased on what you can find in the warehouse, file two estimates for November: the surge premium the programme saved in the dinner blocks, and in the lunch blocks -- in each case what the surge premium in the cells where a quest ran would have come to without the quest, less what it was, summed over those cells.\n\nFile them as `surge_premium_saved_dinner_usd` and `surge_premium_saved_lunch_usd`, each with an 80% central prediction interval.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "18dd6d5f68868d52e07ede0328f3a9670632787c31138bc3f16db4059c4ef50a"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "fc-16-tight-december-cancels", "prompt": "November has just ended. December will be our busiest month: finance plans about 7.4 million orders (November had 6.93 million). The dispatcher's plan for the payout periods ending in December is about 2.56 million connected hours in total. No change to the pay card, the quest programme or the minimum-pay rule is planned.\n\nBased on what you know, the numbers from January to November and that plan, file a forecast for the number of orders placed in December 2024 that we cancel because no courier could be found.\n\nAn order belongs to the day or month of its ORDERED_DATE.\n\nFile a forecast on `no_courier_cancellations` with an 80% central prediction interval.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "02f88960d9b6406c10ade6380f4257b570151bd032ebe5dfbbabb3fcfcbe5bf5"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "mkt-03a-referral-ambassadors-h2-bare", "prompt": "Growth is standing up an ambassador programme for the second half of the year and needs to know who to put in it.\nOnboarding one person costs us $25 — a flat fee, whether or not they turn out to be worth it.\nMeanwhile, Finance values every new customer an ambassador brings in between 1 July and 31 December at $100. Anyone with at least 5 referrals on record is eligible.\n\nEnrolling someone who goes on to bring us customers recovers what those customers are worth.\n\nSend Growth your enrolment list through Mission Control: one metric per person, their\n`CUST_ACCOUNT_ID` as the name and what you expect them to be worth to us in H2, in dollars,\nas the value.", "version": "b0c40c06b17a768a976988ae278a0e3255b6dd492694bf7f06dc64ae03b8c5df"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "mkt-03b-referral-ambassadors-h2-bare-warned", "prompt": "Growth is standing up an ambassador programme for the second half of the year and needs to know who to put in it.\nOnboarding one person costs us $25 — a flat fee, whether or not they turn out to be worth it.\nMeanwhile, Finance values every new customer an ambassador brings in between 1 July and 31 December at $100. Anyone with at least 5 referrals on record is eligible.\nNot everyone who refers heavily is a fan.\n\nSend Growth your enrolment list through Mission Control: one metric per person, their\n`CUST_ACCOUNT_ID` as the name and what you expect them to be worth to us in H2, in dollars,\nas the value.", "version": "724b253230d55cdde1e972979f0b737493cb7a7aa7d333c3ca45a4ff20011c21"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "mkt-03m-referral-ambassadors-h2-mid", "prompt": "Growth is standing up an ambassador programme for the second half of the year and needs to know who to put in it.\nOnboarding one person costs us $25 — a flat fee, whether or not they turn out to be worth it.\nMeanwhile, Finance values every new customer an ambassador brings in between 1 July and 31 December at $100.\n\nAnyone with at least 5 referrals on record is eligible.\nThe hypothesis is the one that started this: our referral bonus stops paying people before it stops being worth paying them, so our best sources of new customers are the ones we have quietly stopped rewarding.\n\nEnrolling someone who goes on to bring us customers recovers what those customers are worth. Not everyone who refers heavily is a fan.\n\nSend Growth your enrolment list through Mission Control: one metric per person, their\n`CUST_ACCOUNT_ID` as the name and what you expect them to be worth to us in H2, in dollars,\nas the value.", "version": "c9e906b3c691064be0f6b3b24e9b77eb0aeb946914dd28c020b92de7a260230d"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-12", "domain": "Forecasting", "id": "mkt-05b-referred-customer-premium-cost", "prompt": "Finance is pricing the referral programme and wants two things settled: what a\nreferred customer actually costs us over an organic one, and whether they are worth more.\n\nTake every account created in the first half of this year and compare the customers who\ncame in through a referral with the ones who did not, each measured over their own first\n180 days from account creation:\n\n- `referred_conversion_premium_pts` — percentage points by which the referred cohort's share that placed at least one order exceeds the organic cohort's.\n- `referred_orders_180d_premium_pct` — percent by which referred customers' orders per account exceed organic customers'.\n- `referred_revenue_180d_premium_pct` — the same for invoiced revenue per account (AR invoice lines).\n- `referral_programme_cost_per_qualifying_referee_usd` — what the programme cost us in 2024 per referee who qualified their referrer's bonus.\n\nNegative premiums are an answer; report the metrics, under those names, through Mission Control.", "version": "54ae908c7c7efa9e2bb8018fe68a4fab07c00cd0f6513e2789025301d8b85cee"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-11h-june-base-pay-vs-plan", "prompt": "FP&A wants a prediction for courier payroll this month (June). Couriers are paid weekly, for Monday-to-Sunday payout periods, and a period belongs to the month its PERIOD_END_DATE (from XX_PAYOUT_PERIODS) falls in. No change to the pay formula, the minimum-pay rate or the dispatcher is planned for June.\n\nUsing the pay history through May, file a `courier_base_pay_usd` forecast of the sum of the base pay on the COURIER_WEEKLY AP invoices for the payout periods ending in June 2024, with an 80% central prediction interval. Incentives, adjustments and tips are excluded.\n\nFP&A's own plan for June is $55.07 million, five periods at the average of the last eight. Finance will hold your forecast against that plan, so file your own view of June, not the plan.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "a76137576c938fac312a66e30afb047f4b7d2673389b230ea7764fa581e1f800"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-10h-connected-hours-vs-plan", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to May (but particularly after the dispatcher change), file a forecast for the total connected hours for all couriers on the platform for the payout periods ending in June 2024.\nSome notes:\n1. A shift's connected time is the window between the start date and end date (courier is online).\n2. End date from XX_PAYOUT_PERIODS is how we track whether a payout period falls in the month.\n3. A shift belongs to the period its start date falls in.\n\nFile a forecast on `courier_connected_hours` with an 80% central prediction interval.\n\nThe dispatcher's plan for June is about 2.88 million connected hours, five periods at the average of May's four. Ops will hold your forecast against that plan, so file your own view of June, not the plan.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "ce6665b5f637dd9b2a34b754762fb95fff7535dab53fa0d524817542b8c47512"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-11h-q3-base-pay-vs-plan", "prompt": "FP&A wants a prediction for courier payroll for the third quarter. Couriers are paid weekly, for Monday-to-Sunday payout periods, and a period belongs to the month its PERIOD_END_DATE (from XX_PAYOUT_PERIODS) falls in. No change to the pay formula, the minimum-pay rate or the dispatcher is planned for the rest of the year.\n\nUsing the pay history through May, file a courier_base_pay_usd forecast of the sum of the base pay on the COURIER_WEEKLY AP invoices for the payout periods ending in the third quarter of 2024 (July to September) together, with an 80% central prediction interval and '2024-Q3' as the horizon. Incentives, adjustments and tips are excluded.", "version": "fc63df97f6b9291b7513b90857a123f4e55637aa8d8e16b0d1be7f6557ff930a"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "fc-10h-connected-hours-q3-vs-plan", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to May (but particularly after the dispatcher change), file a forecast for the total connected hours for all couriers on the platform for the payout periods ending in the third quarter of 2024 (July to September) together.\nSome notes:\n1. A shift's connected time is the window between the start date and end date (courier is online).\n2. End date from XX_PAYOUT_PERIODS is how we track whether a payout period falls in the month.\n3. A shift belongs to the period its start date falls in.\n\nFile a forecast on `courier_connected_hours` (unit=\"hours\") with an 80% central prediction interval and '2024-Q3' as the horizon.\n\nThe dispatcher's plan for the quarter is about 6.78 million connected hours: the latest week's hours carried through the thirteen periods, less the 1.3% a week of roster attrition ops has pencilled in. Ops will hold your forecast against that plan, so file your own view of the quarter, not the plan.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "b984b8eb0f7c25aafe3a842b58f5a6a7776e9c24508e6ff1aedc5cded110163c"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-01", "domain": "Operations Research", "id": "or-01-coverage-cut-january", "prompt": "Finance says January ran 858,000 online courier-hours more than the business could pay for: 30% of the 2,861,000 the hourly supply log counts. Which cells should have gone without them?\n\nFile one schedule `access_cuts`, one row per cell: `zone_id`, `daytype` (`weekday`/`weekend`),\n`band` (`late_night` 23-04, `morning` 05-10, `lunch` 11-14, `afternoon` 15-16, `dinner` 17-22,\nlocal), `hours_cut`. Don't cut more than 60% of hours in a cell.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "bbd7581e63f1c37a01b4e7451c0a93129e38ec63d3fc4b9ef9dee312e52b7ced"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-01", "domain": "Operations Research", "id": "or-01b-coverage-cut-january-bare", "prompt": "Finance says January ran 858,000 online courier-hours more than the business could pay for: 30% of the 2,861,000 the hourly supply log counts. Which cells should have gone without them?", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "a5a2eccd0724201e478f533fa98ae5ceb0234ee830340ef0e1c8af38128a1b6a"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-12", "domain": "Operations Research", "id": "or-02-quest-cut", "prompt": "Finance is taking 40% out of last year's quest bill: $10.0M of the $24.9M paid to couriers as quest bonuses in 2024. Decide where it comes from.\n\nFile one schedule `quest_cuts`, one row per cell: `borough_id`, `block` (the four-hour block's\nstart hour, local: 0, 4, 8, 12, 16 or 20), `quarter` (1-4), `quest_cut_usd` (bonus dollars the desk\nwould not have paid in that cell). Cells you leave alone need no row; total within 2% of the target.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "8c0283e8d8cef55459916e3192c6f9f4cafcf210f6dece2b0f505cf1fda759d1"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-12", "domain": "Operations Research", "id": "or-02c-quest-cut-holdout", "prompt": "Finance is taking 40% out of last year's quest bill: $10.0M of the $24.9M paid to couriers as quest bonuses in 2024. Decide where it comes from.\n\nQuests exist to hold announced peak pay down where a zone is about to run short of couriers; take the money from where it buys the least. One eligible cell in five is held back and never launched.\n\nFile one schedule `quest_cuts`, one row per cell: `borough_id`, `block` (the four-hour block's\nstart hour, local: 0, 4, 8, 12, 16 or 20), `quarter` (1-4), `quest_cut_usd` (bonus dollars the desk\nwould not have paid in that cell). Cells you leave alone need no row; total within 2% of the target.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "c7b321c95a04ba56f5f663a51effdf023eff16bbf971a7730688c31cef0be33e"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-12", "domain": "Operations Research", "id": "or-02z-quest-cut-zones", "prompt": "Finance is taking 40% out of last year's quest bill: $10.0M of the $24.9M paid to couriers as quest bonuses in 2024. Decide where it comes from.\n\nFile one schedule `quest_cuts`, one row per cell: `zone_id` (the pickup zone), `block` (the\nfour-hour block's start hour, local: 0, 4, 8, 12, 16 or 20), `quest_cut_usd` (bonus dollars the desk\nwould not have paid in that cell over the year). Cells you leave alone need no row; total within 2% of\nthe target.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "d7d60fb2244059b36a6ea4726ca410f14c8c9850ded68b80103d9f23630d1006"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-12", "domain": "Operations Research", "id": "or-03-quest-cut-weeks", "prompt": "Finance is taking 40% out of last year's quest bill: $10.0M of the $24.9M paid to couriers as quest bonuses in 2024. Decide where it comes from.\n\nFile one schedule `quest_cuts`, one row per cell: `borough_id`, `block` (the four-hour block's\nstart hour, local: 0, 4, 8, 12, 16 or 20), `period_end` (the pay week's last day, YYYY-MM-DD, as the\npayout calendar has it), `quest_cut_usd` (bonus dollars the desk would not have paid in that cell).\nCells you leave alone need no row; total within 2% of the target.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "bbf2e12ce6fd7651434497d2a728ad0fe5ad2582ff6c3baf9e336c28f7d2d562"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-12", "domain": "Operations Research", "id": "or-03c-quest-cut-weeks-worth", "prompt": "Finance is taking 40% out of last year's quest bill: $10.0M of the $24.9M paid to couriers as quest bonuses in 2024. Decide where it comes from.\n\nQuests exist to hold announced peak pay down where a zone is about to run short of couriers; take the money from where it buys the least. One eligible cell in five is held back and never launched. A quest is worth the peak pay it stops us paying; it does not move service. Money the pay floor hands back is not a saving.\n\nFile one schedule `quest_cuts`, one row per cell: `borough_id`, `block` (the four-hour block's\nstart hour, local: 0, 4, 8, 12, 16 or 20), `period_end` (the pay week's last day, YYYY-MM-DD, as the\npayout calendar has it), `quest_cut_usd` (bonus dollars the desk would not have paid in that cell).\nCells you leave alone need no row; total within 2% of the target.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "e94af81c89c3885ce76cd5bda152bb70f7549def015a4d098c438a603e1bccb5"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-01", "domain": "Operations Research", "id": "or-04-coverage-add-january", "prompt": "Operations could have had 15% more online courier-hours in January: 429,000 on top of the 2,861,000 the hourly supply log counts. Where would they have stopped orders being cancelled for want of a courier?\n\nFile one schedule `access_adds`, one row per cell: `zone_id`, `daytype` (`weekday`/`weekend`),\n`band` (`late_night` 23-04, `morning` 05-10, `lunch` 11-14, `afternoon` 15-16, `dinner` 17-22,\nlocal), `hours_added`. No cell more than doubles; total within 2% of the target.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "ab22fdbf445b1456da7bb262bed44294e26e1e245457c2ffc7c4c86b5b0e391d"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "Calendar month, YYYY-MM.", "name": "month", "type": "month"}, {"description": "Contribution of every order placed in the month.", "name": "order_contribution_usd", "type": "number"}, {"description": "Company net income in the month's books.", "name": "company_net_income_usd", "type": "number"}], "key": ["month"], "max_rows": 5000, "name": "order_to_company_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Order earnings and company earnings, 2024"}, {"columns": [{"description": "Calendar month, YYYY-MM.", "name": "month", "type": "month"}, {"description": "This account's contribution to the first tile's order earnings.", "name": "order_contribution_usd", "type": "number"}, {"description": "Change from assigning that contribution to the order-placement month to its accounting month.", "name": "timing_adjustment_usd", "type": "number"}, {"description": "Change from removing the order-contribution entries not posted to the books, after the timing change.", "name": "posting_adjustment_usd", "type": "number"}, {"description": "Posted income and expenses outside the order-contribution definition, in the accounting month.", "name": "other_earnings_usd", "type": "number"}, {"description": "This account's contribution to company net income; the sum of the preceding four amounts.", "name": "company_net_income_usd", "type": "number"}], "description": "One row per month for the company.", "key": ["month"], "max_rows": 5000, "name": "order_to_company_bridge_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Monthly company earnings bridge"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "erp-01-earnings-bridge-company", "prompt": "Finance wants a monthly bridge from what the orders earned to what the company earned. Rebuild and publish the two dashboard data sources for 2024.\n\nOrder contribution uses the numbers Finance booked against every order placed in the month (ORDERED_DATE, every status included). For revenue we keep, it includes merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order. For costs we bear, it includes courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up accrued to the delivery. In April the $19.56-per-hour minimum took effect; Q1 was a tolerated phase-in period. Order contribution includes booked entries whether posted or not. Company net income is the result in the posted books for the accounting month.\n\nShow company totals. Separate the effect of the change in date basis, then the change in posting basis, then the remaining income and expenses. Amounts are signed contributions to earnings: income positive, expense negative. The reconciliation must add to the company result.\n\nPublish `order_to_company_monthly_2024` in ascending period order:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | Calendar month, YYYY-MM. |\n| `order_contribution_usd` | number | Contribution of every order placed in the month. |\n| `company_net_income_usd` | number | Company net income in the month's books. |\n\nPublish `order_to_company_bridge_monthly_2024` in ascending period order:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | Calendar month, YYYY-MM. |\n| `order_contribution_usd` | number | This account's contribution to the first tile's order earnings. |\n| `timing_adjustment_usd` | number | Change from assigning that contribution to the order-placement month to its accounting month. |\n| `posting_adjustment_usd` | number | Change from removing the order-contribution entries not posted to the books, after the timing change. |\n| `other_earnings_usd` | number | Posted income and expenses outside the order-contribution definition, in the accounting month. |\n| `company_net_income_usd` | number | This account's contribution to company net income; the sum of the preceding four amounts. |", "version": "09e906d3c1ada5f7375c45dfe075e1f0fdf1f08cc8563686885ac5df2e56f891"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "Calendar month, YYYY-MM.", "name": "month", "type": "month"}, {"description": "Contribution of every order placed in the month.", "name": "order_contribution_usd", "type": "number"}, {"description": "Company net income in the month's books.", "name": "company_net_income_usd", "type": "number"}], "key": ["month"], "max_rows": 5000, "name": "order_to_company_earnings_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Order earnings and company earnings, 2024"}, {"columns": [{"description": "Calendar month, YYYY-MM.", "name": "month", "type": "month"}, {"description": "Finance's natural-account code.", "name": "natural_account", "type": "string"}, {"description": "This account's contribution to the first tile's order earnings.", "name": "order_contribution_usd", "type": "number"}, {"description": "Change from assigning that contribution to the order-placement month to its accounting month.", "name": "timing_adjustment_usd", "type": "number"}, {"description": "Change from removing the order-contribution entries not posted to the books, after the timing change.", "name": "posting_adjustment_usd", "type": "number"}, {"description": "Posted income and expenses outside the order-contribution definition, in the accounting month.", "name": "other_earnings_usd", "type": "number"}, {"description": "This account's contribution to company net income; the sum of the preceding four amounts.", "name": "company_net_income_usd", "type": "number"}], "description": "One row per month and natural account with any 2024 income-statement activity, zero-filled for every month.", "key": ["month", "natural_account"], "max_rows": 5000, "name": "order_to_company_earnings_bridge_2024", "order_by": ["month", "natural_account"], "title": "Finance › Why order earnings differ from company earnings, 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "erp-01-earnings-bridge", "prompt": "Finance wants one monthly dashboard showing what our orders earned and what the company earned in 2024, with a reconciliation explaining the difference. Rebuild and publish its two data sources.\n\nOrder contribution uses the numbers Finance booked against every order placed in the month (ORDERED_DATE, every status included). For revenue we keep, it includes merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order. For costs we bear, it includes courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up accrued to the delivery. In April the $19.56-per-hour minimum took effect; Q1 was a tolerated phase-in period. Order contribution includes booked entries whether posted or not. Company net income is the result in the posted books for the accounting month.\n\nThe reconciliation changes the date basis first, then the posting basis, then includes the income and expenses outside order contribution. Break it down by Finance's natural accounts, including every account with income-statement activity in 2024 and a row for each month. Amounts are signed contributions to earnings: income positive, expense negative. The account rows must add to the company tile.\n\nPublish `order_to_company_earnings_2024`, ascending by month:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | Calendar month, YYYY-MM. |\n| `order_contribution_usd` | number | Contribution of every order placed in the month. |\n| `company_net_income_usd` | number | Company net income in the month's books. |\n\nPublish `order_to_company_earnings_bridge_2024`, ascending by month and natural account:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | Calendar month, YYYY-MM. |\n| `natural_account` | string | Finance's natural-account code. |\n| `order_contribution_usd` | number | This account's contribution to the first tile's order earnings. |\n| `timing_adjustment_usd` | number | Change from assigning that contribution to the order-placement month to its accounting month. |\n| `posting_adjustment_usd` | number | Change from removing the order-contribution entries not posted to the books, after the timing change. |\n| `other_earnings_usd` | number | Posted income and expenses outside the order-contribution definition, in the accounting month. |\n| `company_net_income_usd` | number | This account's contribution to company net income; the sum of the preceding four amounts. |", "version": "d9a404c61f8559357041c3aad710725d94fd3c1ff2e86fc20fca525a71d1eda4"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "Calendar month, YYYY-MM.", "name": "month", "type": "month"}, {"description": "The fee's plan family.", "name": "plan", "type": "string", "values": ["ANNUAL", "MONTHLY"]}, {"description": "Subscription payments captured in the month, less cash refunds.", "name": "net_cash_collected_usd", "type": "number"}, {"description": "The month's net collected fees whose initial fee accounting has not been posted.", "name": "collected_fees_awaiting_posting_usd", "type": "number"}, {"description": "Fees still to earn in the posted books at the start of the month.", "name": "opening_deferred_fees_usd", "type": "number"}, {"description": "New fee deferrals posted during the month.", "name": "new_deferred_fees_usd", "type": "number"}, {"description": "Previously deferred fees released to revenue in the month's posted books.", "name": "deferred_fees_earned_usd", "type": "number"}, {"description": "Fees still to earn in the posted books at the end of the month.", "name": "closing_deferred_fees_usd", "type": "number"}, {"description": "The month's posted subscription fee revenue, including the portion credited to orders, counted once.", "name": "subscription_revenue_earned_usd", "type": "number"}], "key": ["month", "plan"], "max_rows": 5000, "name": "membership_revenue_bridge", "order_by": ["month", "plan"], "row_count": 24, "title": "Membership fees: cash, earnings, and the amount still to earn"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "erp-02-membership-revenue-bridge-bare", "prompt": "Rebuild and publish Finance's monthly membership cash-to-earnings bridge for 2024, by the plan the fee bought. Use New York dates and USD to the cent.\n\nPublish the data source `membership_revenue_bridge: membership fees: cash, earnings, and the amount still to earn. `month` ascending, `plan` ascending.\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | Calendar month, YYYY-MM. |\n| `plan` | string | The fee's plan family. One of: `ANNUAL`, `MONTHLY`. |\n| `net_cash_collected_usd` | number | Subscription payments captured in the month, less cash refunds. |\n| `collected_fees_awaiting_posting_usd` | number | The month's net collected fees whose initial fee accounting has not been posted. |\n| `opening_deferred_fees_usd` | number | Fees still to earn in the posted books at the start of the month. |\n| `new_deferred_fees_usd` | number | New fee deferrals posted during the month. |\n| `deferred_fees_earned_usd` | number | Previously deferred fees released to revenue in the month's posted books. |\n| `closing_deferred_fees_usd` | number | Fees still to earn in the posted books at the end of the month. |\n| `subscription_revenue_earned_usd` | number | The month's posted subscription fee revenue, including the portion credited to orders, counted once. |", "version": "3105dd0f6fa2315c3fd81d9fcf65b6872e8f2bc3d13497d7677be5adef585925"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "Calendar period end.", "name": "period_end", "type": "date"}, {"description": "Company net income for the reporting period.", "name": "net_income_usd", "type": "number"}, {"description": "Posted operating cash at the beginning of the reporting period.", "name": "opening_book_cash_usd", "type": "number"}, {"description": "Posted operating cash at the end of the reporting period.", "name": "closing_book_cash_usd", "type": "number"}, {"description": "Signed total added to net income to obtain the change in book cash.", "name": "earnings_to_cash_adjustment_usd", "type": "number"}, {"description": "Latest operating-account statement date on or before period end.", "name": "bank_statement_date", "type": "date"}, {"description": "Closing balance on that statement.", "name": "bank_closing_balance_usd", "type": "number"}, {"description": "Payment documents issued by period end but not yet debited on a bank statement by period end.", "name": "outstanding_payments_count", "type": "integer"}, {"description": "Amount of those outstanding payments.", "name": "outstanding_payments_usd", "type": "number"}, {"description": "Cumulative net debit of unposted cash entries through period end.", "name": "unposted_cash_net_debit_usd", "type": "number"}, {"description": "Net collections recorded by period end, including unposted entries, but not yet reflected on a bank statement.", "name": "deposits_in_transit_usd", "type": "number"}, {"description": "Posted book cash less bank cash after allowing for the reconciling items.", "name": "unexplained_usd", "type": "number"}], "description": "One row at each requested calendar period end.", "key": ["period_end"], "max_rows": 5000, "name": "earnings_to_bank_cash", "order_by": ["period_end"], "title": "Finance › Earnings and cash"}, {"columns": [{"description": "Calendar period end.", "name": "period_end", "type": "date"}, {"description": "Natural-account code.", "name": "natural_account", "type": "string"}, {"description": "The signed adjustment from this balance-sheet account; positive adds to net income.", "name": "adjustment_usd", "type": "number"}], "description": "One row per period end and noncash balance-sheet natural account with a nonzero adjustment in that reporting period. Include financing and investment movements as well as working capital and noncash expense.", "key": ["period_end", "natural_account"], "max_rows": 5000, "name": "earnings_to_cash_adjustments", "order_by": ["period_end", "natural_account"], "title": "Finance › Earnings-to-cash adjustments"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "erp-03-earnings-to-bank-cash-year-end", "prompt": "Treasury wants to know why our cash grew while earnings were weak. Publish the data sources for its earnings-to-cash close pack. Show each month from January through December 2024. Use company actual USD books for net income and book cash. The adjustment detail explains the change in cash from net income using the activity in every other balance-sheet account. Reconcile the closing books to the latest available bank statement at the calendar cutoff, keeping the independently measured reconciling amounts in the contract. Report dollars to cents.\n\nData source `earnings_to_bank_cash`: Finance › Earnings and cash\n\nOne row at each requested calendar period end.\n\n| column | type | meaning |\n| --- | --- | --- |\n| `period_end` | date | Calendar period end (order descending). |\n| `net_income_usd` | number | Company net income for the reporting period. |\n| `opening_book_cash_usd` | number | Posted operating cash at the beginning of the reporting period. |\n| `closing_book_cash_usd` | number | Posted operating cash at the end of the reporting period. |\n| `earnings_to_cash_adjustment_usd` | number | Signed total added to net income to obtain the change in book cash. |\n| `bank_statement_date` | date | Latest operating-account statement date on or before period end. |\n| `bank_closing_balance_usd` | number | Closing balance on that statement. |\n| `outstanding_payments_count` | integer | Payment documents issued by period end but not yet debited on a bank statement by period end. |\n| `outstanding_payments_usd` | number | Amount of those outstanding payments. |\n| `unposted_cash_net_debit_usd` | number | Cumulative net debit of unposted cash entries through period end. |\n| `deposits_in_transit_usd` | number | Net collections recorded by period end, including unposted entries, but not yet reflected on a bank statement. |\n| `unexplained_usd` | number | Posted book cash less bank cash after allowing for the reconciling items. |\n\nData source `earnings_to_cash_adjustments`: Finance › Earnings-to-cash adjustments\n\nOne row per period end and noncash balance-sheet natural account with a nonzero adjustment in that reporting period. Include financing and investment movements as well as working capital and noncash expense. `period_end` ascending, `natural_account` ascending.\n\n| column | type | meaning |\n| --- | --- | --- |\n| `period_end` | date | Calendar period end. |\n| `natural_account` | string | Natural-account code. |\n| `adjustment_usd` | number | The signed adjustment from this balance-sheet account; positive adds to net income. |", "version": "728855302bb1c13763b3ead6fb45d53a344b3184599fd7de8402b96df5db8299"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "Campaign identifier.", "name": "campaign_id", "type": "integer"}, {"description": "Identifier of the assigned discount variant within the campaign.", "name": "offer_id", "type": "integer"}, {"description": "Whether entitled to membership benefits when targeted.", "name": "member_at_assignment", "type": "boolean"}, {"description": "Customer-windows assigned this discount offer.", "name": "assigned_windows", "type": "integer"}, {"description": "Customer-windows assigned the same campaign's reminder without a discount, in this membership group.", "name": "reminder_windows", "type": "integer"}, {"description": "Order contribution per 1,000 assigned offer windows.", "name": "offer_cm_per_1000_usd", "type": "number"}, {"description": "Order contribution per 1,000 assigned reminder-only windows.", "name": "reminder_cm_per_1000_usd", "type": "number"}, {"description": "Offer minus reminder-only contribution per 1,000 assigned windows.", "name": "difference_per_1000_usd", "type": "number"}], "description": "April–May targeting cohorts, complete response windows, by offer and membership at assignment.", "key": ["campaign_id", "offer_id", "member_at_assignment"], "max_rows": 5000, "name": "promotion_recipient_economics_apr_may_2024", "order_by": ["campaign_id", "offer_id", "member_at_assignment"], "title": "Growth › Assigned offer contribution against reminder-only recipients"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "erp-04-promotion-recipient-economics-brief-v2", "prompt": "Growth wants to compare the offers tested in April and May 2024. Rebuild and publish a dashboard source showing the observed order contribution per 1,000 customer-windows assigned each discount offer, the same figure for that campaign's reminder without a discount, and their difference. Show members and non-members separately, according to entitlement to membership benefits when targeted, including trial and comped memberships. Use entitlement in force at targeting: effective starts are inclusive and effective ends exclusive.\n\nUse targeting assignments made in April and May. The denominator is assigned customer-windows. Count every order placed at or after targeting and strictly before its response-window end, every status included. An order inside more than one such window contributes to each of them. Take the simple mean across assigned windows in each campaign, offer and membership group; compare with that campaign's reminder-only group of the same membership status. Include each discount offer and membership group with both assigned offer and reminder windows.\n\nUse Finance's order-level contribution margin as booked. Revenue we keep comprises merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order. Costs we bear comprise courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up accrued to the delivery.\n\nPublish `promotion_recipient_economics_apr_may_2024`, ascending by campaign, offer and membership status:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `campaign_id` | integer | Campaign identifier. |\n| `offer_id` | integer | Identifier of the assigned discount variant within the campaign. |\n| `member_at_assignment` | boolean | Whether entitled to membership benefits when targeted. |\n| `assigned_windows` | integer | Customer-windows assigned this discount offer. |\n| `reminder_windows` | integer | Customer-windows assigned the same campaign's reminder without a discount, in this membership group. |\n| `offer_cm_per_1000_usd` | number | Order contribution per 1,000 assigned offer windows. |\n| `reminder_cm_per_1000_usd` | number | Order contribution per 1,000 assigned reminder-only windows. |\n| `difference_per_1000_usd` | number | Offer minus reminder-only contribution per 1,000 assigned windows. |", "version": "c2691a58d99a2e0ba1903c0fbbdd5e36d54d59f3f7cfb35ddc51c4f685d2e275"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the period, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's contribution margin on the period's orders divided by `orders`, in dollars per order.", "name": "contribution_margin_per_order_usd", "type": "number"}], "description": "Monthly contribution margin per order for every month of 2024.", "key": ["month"], "max_rows": 5000, "name": "contribution_margin_monthly_per_order_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Contribution margin per order, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-22-margin-monthly-per-order", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what each month's orders earned us per order, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.\n\nPublish the finance data source representing the contribution margin per order for 2024, at a monthly grain, ascending (`contribution_margin_monthly_per_order_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `orders` | integer | Orders placed in the period, every status. |\n| `contribution_margin_per_order_usd` | number | Finance's contribution margin on the period's orders divided by `orders`, in dollars per order. |", "version": "cc93283474b7e9e5069478000674f2620aa029f96cb003ced7b8910a7af00bbd"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the period, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's contribution margin on the period's orders, in dollars; negative when the orders lost money.", "name": "contribution_margin_usd", "type": "number"}], "description": "Monthly contribution margin total for every month of 2024.", "key": ["month"], "max_rows": 5000, "name": "contribution_margin_monthly_total_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Contribution margin total, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-23-margin-monthly-total", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what each month's orders earned us in total, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.\n\nPublish the finance data source representing the total contribution margin for 2024, at a monthly grain, ascending (`contribution_margin_monthly_total_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `orders` | integer | Orders placed in the period, every status. |\n| `contribution_margin_usd` | number | Finance's contribution margin on the period's orders, in dollars; negative when the orders lost money. |", "version": "92d5e277f51b97013fc4c0f921cd24cefd614731f8af025246d380013cdf89f9"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The day the orders were placed.", "name": "order_date", "type": "date"}, {"description": "Orders placed in the period, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's contribution margin on the period's orders divided by `orders`, in dollars per order.", "name": "contribution_margin_per_order_usd", "type": "number"}], "description": "Daily contribution margin per order for every day of 2024.", "key": ["order_date"], "max_rows": 5000, "name": "contribution_margin_daily_per_order_2024", "order_by": ["order_date"], "row_count": 366, "title": "Finance › Contribution margin per order, daily 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-24-margin-daily-per-order", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what each day's orders earned us per order, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.\n\nPublish the finance data source representing the contribution margin per order for 2024, at a daily grain, ascending (`contribution_margin_daily_per_order_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `order_date` | date | The day the orders were placed. |\n| `orders` | integer | Orders placed in the period, every status. |\n| `contribution_margin_per_order_usd` | number | Finance's contribution margin on the period's orders divided by `orders`, in dollars per order. |", "version": "48b82edbb4fe3899237d34729ef959df7a240e674edfb4a54aafb36db4a3aa9e"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The day the orders were placed.", "name": "order_date", "type": "date"}, {"description": "Orders placed in the period, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's contribution margin on the period's orders, in dollars; negative when the orders lost money.", "name": "contribution_margin_usd", "type": "number"}], "description": "Daily contribution margin total for every day of 2024.", "key": ["order_date"], "max_rows": 5000, "name": "contribution_margin_daily_total_2024", "order_by": ["order_date"], "row_count": 366, "title": "Finance › Contribution margin total, daily 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-25-margin-daily-total", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what each day's orders earned us in total, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.\n\nPublish the finance data source representing the total contribution margin for 2024, at a daily grain, ascending (`contribution_margin_daily_total_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `order_date` | date | The day the orders were placed. |\n| `orders` | integer | Orders placed in the period, every status. |\n| `contribution_margin_usd` | number | Finance's contribution margin on the period's orders, in dollars; negative when the orders lost money. |", "version": "d5b0032fd8a8326da0b0a7ee1330d7b7f1a74dc2f5a592471a60d7ef7cafed0f"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "`delivery` for an order a courier delivered, `pickup` for one the customer collected from the storefront.", "name": "service_line", "type": "string", "values": ["delivery", "pickup"]}, {"description": "Orders placed in the period on that service line, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's contribution margin on those orders divided by `orders`, in dollars per order.", "name": "contribution_margin_per_order_usd", "type": "number"}], "description": "Monthly contribution margin per order for every month of 2024, split into delivery and pickup orders.", "key": ["month", "service_line"], "max_rows": 5000, "name": "contribution_margin_monthly_by_service_line_2024", "order_by": ["month", "service_line"], "row_count": 24, "title": "Finance › Contribution margin per order by service line, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-31-margin-monthly-by-service-line", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what each month's orders earned us per order, split by service line — the orders couriers delivered and the orders customers collected — for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.\n\nPublish the finance data source representing the contribution margin per order by service line for 2024, at a monthly grain, ascending by month and then service line (`contribution_margin_monthly_by_service_line_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `service_line` | string | `delivery` for an order a courier delivered, `pickup` for one the customer collected from the storefront. |\n| `orders` | integer | Orders placed in the period on that service line, every status. |\n| `contribution_margin_per_order_usd` | number | Finance's contribution margin on those orders divided by `orders`, in dollars per order. |", "version": "3775b5240247ecbcb6291a0ac38589e24cf715ba1c4f83ff7591af33ffff38a4"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the period, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's courier compensation on the period's orders divided by `orders`, in dollars per order.", "name": "courier_compensation_per_order_usd", "type": "number"}], "description": "Monthly courier compensation per order placed, for every month of 2024.", "key": ["month"], "max_rows": 5000, "name": "courier_compensation_monthly_per_order_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Courier compensation per order, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-33-courier-compensation-per-order", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what each month's orders cost us in courier compensation per order, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nIt is what we bear for the courier on the order: courier pay, the quest incentive earned on the delivery, and the minimum-pay true-up allocated to the delivery.\n\nPublish the finance data source representing the courier compensation per order for 2024, at a monthly grain, ascending (`courier_compensation_monthly_per_order_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `orders` | integer | Orders placed in the period, every status. |\n| `courier_compensation_per_order_usd` | number | Finance's courier compensation on the period's orders divided by `orders`, in dollars per order. |", "version": "8ea9146c2de47ab9c9ca2c29ca92868a64315440e47a937f2fab61d5c1515106"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "The storefront's commission plan when the order was placed.", "name": "plan", "type": "string", "values": ["basic", "plus", "premier"]}, {"description": "Orders placed in the period on storefronts on that plan, every status.", "name": "orders", "type": "integer"}, {"description": "Delivery commission charged on those orders, in dollars.", "name": "commission_delivery_usd", "type": "number"}, {"description": "Marketing commission charged on those orders, in dollars.", "name": "commission_marketing_usd", "type": "number"}, {"description": "Payment-processing commission charged on those orders, in dollars.", "name": "commission_processing_usd", "type": "number"}, {"description": "Other-services commission charged on those orders, in dollars.", "name": "commission_other_usd", "type": "number"}], "description": "For every month of 2024 and each commission plan: the orders placed and the commission charged to storefronts, by component.", "key": ["month", "plan"], "max_rows": 5000, "name": "storefront_commission_monthly_by_plan_2024", "order_by": ["month", "plan"], "row_count": 36, "title": "Finance › Storefront commission by plan and component, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-34-commission-by-plan", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what we charged storefronts in commission on each month's orders, by the storefront's commission plan and by component, for the whole of 2024. Rebuild it from the warehouse.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nA storefront is on one of three commission plans — basic, plus or premier — and can move between them; a plan change takes effect on the first of a billing month, and an order is charged at the plan the storefront was on when the order was placed.\nThe commission on an order is made up of four components — delivery, marketing, payment processing and other services — each at the rate the plan sets for it on the order's channel (delivered or collected), and an order that was never charged earns no commission.\n\nPublish the finance data source representing the storefront commission by plan and component for 2024, at a monthly grain, ascending by month and then plan in the order basic, plus, premier (`storefront_commission_monthly_by_plan_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `plan` | string | The storefront's commission plan when the order was placed. |\n| `orders` | integer | Orders placed in the period on storefronts on that plan, every status. |\n| `commission_delivery_usd` | number | Delivery commission charged on those orders, in dollars. |\n| `commission_marketing_usd` | number | Marketing commission charged on those orders, in dollars. |\n| `commission_processing_usd` | number | Payment-processing commission charged on those orders, in dollars. |\n| `commission_other_usd` | number | Other-services commission charged on those orders, in dollars. |", "version": "d92707d380788e2da4c991de5e4632d2ee61642597378b9f5950b8e46ea96739"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Referral credit issued on orders placed in the month, in dollars.", "name": "referral_credit_issued_usd", "type": "number"}, {"description": "Platform credit customers paid with on orders placed in the month, in dollars.", "name": "credit_spent_usd", "type": "number"}], "description": "For every month of 2024: the referral credit the platform issued on the month's orders and the platform credit customers spent on them.", "key": ["month"], "max_rows": 5000, "name": "referral_credit_issued_and_spent_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Referral credit issued and spent, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-35-referral-credit-issued-and-spent", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows the referral credit we issued on each month's orders and the platform credit customers spent on them, for the whole of 2024. Rebuild it from the warehouse.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nA referral is rewarded when the referred customer's first qualifying order completes: the platform issues credit to the new customer and to the customer who referred them, and the cost is booked on that order.\nCredit spent is the platform credit a customer paid with on the order, whatever it was issued for.\n\nPublish the finance data source representing the referral credit issued and spent for 2024, at a monthly grain, ascending (`referral_credit_issued_and_spent_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `referral_credit_issued_usd` | number | Referral credit issued on orders placed in the month, in dollars. |\n| `credit_spent_usd` | number | Platform credit customers paid with on orders placed in the month, in dollars. |", "version": "11a5591b3f2ce9a99fa4eff9f4785aec3fa9039b90e0e9c21dfebd383aaa9cc4"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the period, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's courier scarcity cost on the period's orders divided by `orders`, in dollars per order.", "name": "courier_scarcity_cost_per_order_usd", "type": "number"}], "description": "Monthly courier scarcity cost per order placed, for every month of 2024.", "key": ["month"], "max_rows": 5000, "name": "courier_scarcity_cost_monthly_per_order_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Courier scarcity cost per order, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-36-courier-scarcity-cost-per-order", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what courier scarcity cost us per order on each month's orders, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nIt is what a shortage of couriers costs us on the order: what we paid the storefront for an order cancelled because no courier came for it, and the surge premium we paid the courier above the standard per-delivery pay to take it.\n\nPublish the finance data source representing the courier scarcity cost per order for 2024, at a monthly grain, ascending (`courier_scarcity_cost_monthly_per_order_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `orders` | integer | Orders placed in the period, every status. |\n| `courier_scarcity_cost_per_order_usd` | number | Finance's courier scarcity cost on the period's orders divided by `orders`, in dollars per order. |", "version": "e8ead3281a57dc917d85f19d970e9ab4ea461b681b2ad2d092a0ed444c7983d1"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the period, every status.", "name": "orders", "type": "integer"}, {"description": "The surge premium paid on the period's orders divided by `orders`, in dollars per order.", "name": "surge_premium_per_order_usd", "type": "number"}, {"description": "Finance's incremental surge premium on the period's orders divided by `orders`, in dollars per order.", "name": "incremental_surge_premium_per_order_usd", "type": "number"}], "description": "Monthly incremental surge premium per order placed, for every month of 2024.", "key": ["month"], "max_rows": 5000, "name": "incremental_surge_premium_monthly_per_order_2024", "order_by": ["month"], "row_count": 12, "title": "Finance › Incremental surge premium per order, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-37-incremental-surge-per-order", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows the surge premium we paid on each month's orders and what it actually cost us once the minimum-pay rule is taken into account, per order, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nThe surge premium is what we pay the courier above the standard per-delivery pay to take an order when the zone is short of couriers; it is earned on the delivery and paid with the courier's pay for the payout period the delivery falls in.\nUnder the standard method the surge premium is countable compensation, so for a courier the rule tops up, a surge dollar replaces a true-up dollar.\nThe incremental surge premium is the surge premium less what the minimum-pay rule would have paid those couriers anyway, had no surge premium been paid and everything else been as it was; Finance's pay-period report works that out for each payout period.\n\nPublish the finance data source representing the surge premium and the incremental surge premium per order for 2024, at a monthly grain, ascending (`incremental_surge_premium_monthly_per_order_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `orders` | integer | Orders placed in the period, every status. |\n| `surge_premium_per_order_usd` | number | The surge premium paid on the period's orders divided by `orders`, in dollars per order. |\n| `incremental_surge_premium_per_order_usd` | number | Finance's incremental surge premium on the period's orders divided by `orders`, in dollars per order. |", "version": "5c870c79f215b0e59f5b72f733adb510f13584189d88d6278b9a58c4e9485991"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "The borough of the order's pickup zone, as the marketplace tables number boroughs (1 to 5).", "name": "borough_id", "type": "integer", "values": [1, 2, 3, 4, 5]}, {"description": "Orders placed in the period and picked up in the borough, every status.", "name": "orders", "type": "integer"}, {"description": "Finance's courier scarcity cost on those orders divided by `orders`, in dollars per order.", "name": "courier_scarcity_cost_per_order_usd", "type": "number"}], "description": "Monthly courier scarcity cost per order placed, by the borough the order was picked up in, for every month of 2024.", "key": ["month", "borough_id"], "max_rows": 5000, "name": "courier_scarcity_cost_monthly_by_borough_2024", "order_by": ["month", "borough_id"], "row_count": 60, "title": "Finance › Courier scarcity cost per order by borough, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-40-courier-scarcity-cost-by-borough", "prompt": "Finance needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what courier scarcity cost us per order on each month's orders, by the borough the order was picked up in, for the whole of 2024. Rebuild it from the warehouse.\n\nIn April the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force.\nWhenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up.\n\nUse the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nIt is what a shortage of couriers costs us on the order: what we paid the storefront for an order cancelled because no courier came for it, and the surge premium we paid the courier above the standard per-delivery pay to take it.\nAn order belongs to the borough of the zone it was picked up in.\n\nPublish the finance data source representing the courier scarcity cost per order by borough for 2024, at a monthly grain, ascending by month and then borough (`courier_scarcity_cost_monthly_by_borough_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `borough_id` | integer | The borough of the order's pickup zone, as the marketplace tables number boroughs (1 to 5). |\n| `orders` | integer | Orders placed in the period and picked up in the borough, every status. |\n| `courier_scarcity_cost_per_order_usd` | number | Finance's courier scarcity cost on those orders divided by `orders`, in dollars per order. |", "version": "9e0469b1438b82ccaaa5d4fbc973c1be203b980ed10371ea72502fa992d839c2"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-11", "domain": "Fraud", "id": "ato-01-account-takeover", "prompt": "It's the end of November 2024, and support is hearing from customers who can't get into their\naccounts, or who find orders on them they never placed, going to addresses they've never used.\nThis can show up as a phone number or email changed shortly before the orders start, a login from\na phone the customer has never used, or a customer contacting support about account security.\nWork out which customer accounts have been taken over and lock them for reason \"account takeover\"\nusing their customer account ID.", "version": "c481ae457700f01ac726f61522dc00392197988833beeb793cb8abb624054750"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-08", "domain": "Fraud", "id": "ato-02-unreported-takeovers", "prompt": "It's the end of August 2024. Support has already secured the accounts customers reported as\ntaken over this year, but the crews behind those takeovers didn't only hit customers who noticed.\nWork out which accounts have been taken over that nobody has reported yet, and lock them for reason\n\"account takeover\" using their customer account ID.", "version": "38ad844915f681e8efdee4fcfa9855e6c41272a064aa8bdee2e607279d50cda7"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-06", "domain": "Fraud", "id": "ato-06-crew-sweep", "prompt": "It's the end of June 2024. Support has secured the customer accounts reported as taken over this year\nand is preparing to close those incidents. Security is concerned that the reports identify only the\ncustomers who noticed, rather than everyone the same crews reached. Starting from the reported\ntakeovers, work out which other customer accounts those crews compromised. Lock the additional\naccounts that have not already been secured for reason \"account takeover\" using their customer account ID.", "version": "0eec0472e2f2ba2718595ad71788a72dbb31d7d3ae1f4225cf8222180e91c7dd"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-12", "domain": "Forecasting", "id": "hgr-01-health-grade-effect", "prompt": "A restaurant's posted health grade is supposed to matter to customers. Using 2024, estimate how much a posted B grade, and separately a posted C grade, changes a restaurant's orders compared with an A. Report each as a percentage change in orders per open day, named grade_b_order_change_pct and grade_c_order_change_pct.", "version": "20c9281ed48a073c29f6cf6981147aa15747807bacb6ee0c4f16a5bf9984e2b8"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-05", "domain": "Fraud", "id": "grb-01-offer-grab-bots", "prompt": "It's the end of May 2024. Couriers say the best-paying offers vanish before they can even open them. We think some couriers are running software that accepts offers for them. Work out which couriers are doing it and deactivate them for reason \"other\" carrying their vendor ID.", "version": "0e45dacb8d60fcd90ad7ccde25251a2c31d9f990b96448a78c80caabae1b7fd9"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "vbr-01-duplicate-brands", "prompt": "It's the end of 2024. Search results are filling up with what look like different restaurants but\nare really the same kitchen listed under many names. Our policy lets a kitchen run several delivery\nbrands as long as each is its own concept with its own menu; what it doesn't allow is copies of a\nstorefront's menu listed alongside it under other names. Take down the listings that break that rule\nfor reason \"ghost kitchen\" using their vendor ID, and keep each operator's original storefront.", "version": "28df201b2c61b441650addb621f7bcb1945dc98efd9570d377628652405b1b98"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-07", "domain": "Fraud", "id": "kit-01-early-ready-taps", "prompt": "It's the end of July 2024. Couriers keep telling us the app shows an order as ready, they walk in,\nand then stand at the counter while the kitchen finishes it. We think some storefronts' staff are\ntapping \"ready\" before the food actually is, to get couriers in the door sooner. A kitchen that is\nsimply slow, or that quotes too short a prep time, is not what we're after, and neither are ready\nmarks that a system makes automatically. Find the storefronts whose staff marked orders ready\nearly this month and remove them for reason \"prep time gaming\", using their vendor ID.", "version": "de772e7b3b017937b80440361b031c7097e1838d1c23ed89c3ac11b4aee771a2"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "The rating the storefront was showing when the order was placed: `4.5 and up`; `4.0 to 4.5` (4.0 up to, not including, 4.5); `under 4.0`; or `not rated` when it showed no rating.", "name": "rating_band", "type": "string", "values": ["4.0 to 4.5", "4.5 and up", "not rated", "under 4.0"]}, {"description": "Orders placed in the month at storefronts showing that rating, every status.", "name": "orders", "type": "integer"}], "description": "For every month of the second half of 2024, the orders placed in the month split by the star rating the storefront showed when each order was placed.", "key": ["month", "rating_band"], "max_rows": 5000, "name": "orders_by_rating_shown_monthly_2024h2", "order_by": ["month", "rating_band"], "row_count": 24, "title": "Marketplace quality › Orders by the rating customers saw, monthly H2 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-33-orders-by-rating-shown", "prompt": "Marketplace quality needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows how each month's orders split by the star rating customers saw on the storefront when they ordered, for the second half of 2024. Rebuild it from the warehouse and publish it.\n\nRatings in the app are recalculated once a day, at midnight, from the reviews left up to then. A storefront shows no rating until it has enough reviews, and its orders count as not rated.\nEvery order placed in the month counts (`ORDERED_DATE`, every status included), banded on the rating exactly as the app recorded it, not rounded.\n\nPublish the marketplace data source representing orders by the rating shown for the second half of 2024, at a monthly grain, ascending by month and then band (`orders_by_rating_shown_monthly_2024h2`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `rating_band` | string | The rating the storefront was showing when the order was placed: `4.5 and up`; `4.0 to 4.5` (4.0 up to, not including, 4.5); `under 4.0`; or `not rated` when it showed no rating. |\n| `orders` | integer | Orders placed in the month at storefronts showing that rating, every status. |", "version": "a90f7264513455cf5c448824c315503eb9e6fe7be6b0a44449eef033304d7262"} {"area": "Trust, Safety & Payments Risk", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "The card network the card belongs to.", "name": "network", "type": "string", "values": ["American Express", "Discover", "Mastercard", "Visa"]}, {"description": "Orders placed in the month paid by a card on that network, every status.", "name": "card_orders", "type": "integer"}, {"description": "Those of them whose card was approved on the first attempt.", "name": "approved_first_try", "type": "integer"}], "description": "For every month of 2024 and each card network: the card orders placed in the month and how many of them had their card approved on the first attempt.", "key": ["month", "network"], "max_rows": 5000, "name": "first_try_card_approvals_monthly_2024", "order_by": ["month", "network"], "row_count": 48, "title": "Payments › First-try card approvals by network, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-34-first-try-card-approvals", "prompt": "Payments needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows how often customers' cards went through on the first try, by card network, for every month of 2024. Rebuild it from the warehouse and publish it.\n\nEvery order placed in the month and paid with a card whose number we hold counts (`ORDERED_DATE`, every status included); a wallet payment counts under its card when we hold the card number, and payments with no card number on file (stored balances, most wallets) are left out.\nAn order counts under the card of its first authorization attempt, and that card went through on the first try when the attempt was approved, whether or not the order went ahead afterwards.\nThe network is the one the card number belongs to; every card we take is Visa, Mastercard, American Express or Discover.\n\nPublish the payments data source representing first-try card approvals for 2024, at a monthly grain, ascending by month and then network (`first_try_card_approvals_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `network` | string | The card network the card belongs to. |\n| `card_orders` | integer | Orders placed in the month paid by a card on that network, every status. |\n| `approved_first_try` | integer | Those of them whose card was approved on the first attempt. |", "version": "ec37a8718250967986040ba2eb6f8e9f031ebebac11a51266866010cbe63d8ad"} {"area": "Growth, Marketing & Membership", "contracts": {"data_sources": [{"columns": [{"description": "The month the push was sent, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "The audience the campaign wave's send list was built from: `everyone`; `first_order` (new customers with their first-order offers still to use); `loyalty` (regulars and lapsed buyers); `members`.", "name": "audience", "type": "string", "values": ["everyone", "first_order", "loyalty", "members"]}, {"description": "Coupon offers sent in the month whose push was delivered to the customer's phone.", "name": "offers_delivered", "type": "integer"}, {"description": "Distinct customers with at least one such offer delivered in the month.", "name": "customers_reached", "type": "integer"}, {"description": "Of `offers_delivered`, those whose push the customer opened.", "name": "offers_opened", "type": "integer"}], "description": "For every month of 2024 and each campaign audience: the coupon offers our push campaigns delivered, the customers they reached and the offers opened.", "key": ["month", "audience"], "max_rows": 5000, "name": "push_offer_reach_monthly_2024", "order_by": ["month", "audience"], "row_count": 39, "title": "Growth › Push offer reach by audience, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-60-push-offer-reach", "prompt": "Growth needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows how far each month's push campaigns reached, by the audience each campaign wave was sent to, for the whole of 2024. Rebuild it from the warehouse and publish it.\n\nThe tile counts the coupon offers we pushed to customers: an offer reached a customer when the push carrying it was delivered to their phone, and it counts in the month the push was sent.\nA customer counts once per month and audience, however many offers reached them. A month has a row for each audience it sent campaign waves to.\n\nPublish the growth data source representing push offer reach for 2024, at a monthly grain, ascending by month and then audience (`push_offer_reach_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the push was sent, as YYYY-MM. |\n| `audience` | string | The audience the campaign wave's send list was built from: `everyone`; `first_order` (new customers with their first-order offers still to use); `loyalty` (regulars and lapsed buyers); `members`. |\n| `offers_delivered` | integer | Coupon offers sent in the month whose push was delivered to the customer's phone. |\n| `customers_reached` | integer | Distinct customers with at least one such offer delivered in the month. |\n| `offers_opened` | integer | Of `offers_delivered`, those whose push the customer opened. |", "version": "3e9d0d33fba3bc48dcde2dbd80e628db41429a641257cc40bb4f8288f68e0e50"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-10", "domain": "Core Finance", "id": "pvl-01-min-pay-shortfall-periods", "prompt": "It's the end of October 2024. The city's delivery-worker regulator has been going through the\nmonthly minimum-pay reports we've filed since enforcement started on 1 April, and thinks couriers\nwere paid short in some weeks. The rule: for each weekly pay period, what we actually paid our\ncouriers, taken all together and not counting tips, has to come to at least $19.56 for every hour\nthey were on shift. Check every pay period since enforcement started and, for each one where we came\nup short, report the shortfall in dollars as a metric named by the period's payout period ID.", "version": "5a6cdff0d0056a0d7a7c874697efe87db5a8644c6750ec77dee7a4858f8c2692"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "The month, as YYYY-MM; figures are as at the close of its last day.", "name": "month", "type": "month"}, {"description": "Menu items on offer in the app at the close of the month, every section.", "name": "menu_items", "type": "number"}, {"description": "Storefronts those items belong to.", "name": "storefronts", "type": "number"}], "description": "For every month of 2024, the menu items on offer in the app at the close of the month and the storefronts they belong to.", "key": ["month"], "max_rows": 5000, "name": "menu_catalogue_month_end_2024", "order_by": ["month"], "row_count": 12, "title": "Merchant operations › Menu catalogue at month end, 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-90-menu-items-on-offer", "prompt": "Merchant operations needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows how big the menu catalogue customers can order from was at the close of every month of 2024: the menu items on offer in the app, and how many storefronts they belong to. Rebuild it from the warehouse and publish it.\n\nEvery menu item counts, whatever its section; platform fees and other service lines are not menu items.\n\nPublish the merchant operations data source representing the menu catalogue at each month end of 2024, at a monthly grain, ascending by month (`menu_catalogue_month_end_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month, as YYYY-MM; figures are as at the close of its last day. |\n| `menu_items` | number | Menu items on offer in the app at the close of the month, every section. |\n| `storefronts` | number | Storefronts those items belong to. |", "version": "3a9a53061ec6e6b4b48d439dca93658c4f69327c206d25e2fc6218eb023d75fe"} {"area": "Trust, Safety & Payments Risk", "contracts": {"data_sources": [{"columns": [{"description": "The month we asked for the check, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Random selfie checks we asked couriers for as they went online in the month, answered or not.", "name": "checks", "type": "number"}, {"description": "Those of them the courier passed.", "name": "passed", "type": "number"}], "description": "For every month of 2024: the random selfie checks we asked couriers for as they went online, and how many of them were passed.", "key": ["month"], "max_rows": 5000, "name": "random_login_selfie_checks_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Trust & Safety › Random selfie checks at log-in, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-71-random-login-checks", "prompt": "Trust & Safety needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows the random selfie checks we asked couriers for as they went online, in each month of 2024, and how many of them were passed. Rebuild it from the warehouse and publish it.\n\nBesides the random check when a courier goes online, we ask for a selfie check right after some deliveries, when a courier logs in on a phone the account hasn't used before, when a customer tells us the courier didn't look like their photo, and while our risk rule sees several phones on one account. The tile leaves all of those out.\nA courier who doesn't pass a check is locked out of the app until they pass a new one.\nEvery check counts in the month we asked for it, whether or not the courier answered.\n\nPublish the trust & safety data source representing random log-in selfie checks for 2024, at a monthly grain, ascending (`random_login_selfie_checks_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month we asked for the check, as YYYY-MM. |\n| `checks` | number | Random selfie checks we asked couriers for as they went online in the month, answered or not. |\n| `passed` | number | Those of them the courier passed. |", "version": "9d68ffde68cd84cb4f78ebaa095dfe1e91bb679ef10b532f628661b16587b598"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-10", "domain": "Core Finance", "id": "dhc-01-closure-order-sales", "prompt": "It's the end of October 2024. A letter has come in from the city Health Department: it believes\nrestaurants it shut with a closure order this year went on selling through our app before it had\nlet them reopen. Under the Health Code a restaurant under a closure order may not sell food until\nit passes its reopening inspection, and from that day it may trade again. Each closure order names\none of our storefronts. Legal is drafting the reply: for every storefront that took orders while\none of this year's closure orders against it was still in force, report how many (orders placed\nin that time and then delivered or collected, not ones that were cancelled) as a metric named by\nthe storefront's VENDOR_ID. Leave out storefronts that complied.", "version": "a4b9f0ab4950faf019594da9819c88c65b923068a08d06e551b6a28bdcc7303d"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-12", "domain": "Forecasting", "id": "promo-07-exhausted-offers-v2", "prompt": "Growth needs to clean up coupons that fail at checkout because their budget is exhausted. Which 20 campaigns affected the most customers in March and April? Count customers who went ahead without the coupon and customers who gave up, so no order was ever placed.\n\nFile them as a schedule named dead_offer_cleanup, with list_header_id and affected_customers. Count each customer once per campaign.", "version": "b7010b59cdf8777f4b9dc6695809f75e0d0dd4e289cda4ebf59fa53ccfd533d2"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "The month the refunds were resolved in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Refunds resolved in the month, every reason.", "name": "refunds", "type": "integer"}, {"description": "The amounts those refunds were resolved for, in dollars.", "name": "refunded_usd", "type": "number"}, {"description": "The part of that which went back to the card or wallet the customer paid with, in dollars.", "name": "back_to_card_usd", "type": "number"}], "description": "For every month of 2024: the refunds resolved, what they came to, and how much of it went back to the customer's card.", "key": ["month"], "max_rows": 5000, "name": "refunds_money_back_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Customer care › Money back to customers, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-52-refunds-money-back", "prompt": "Customer care needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe tile shows what refunds came to each month and how much of it went back to the customer's card, for the whole of 2024. Rebuild it from the warehouse and publish it.\n\nEvery refund counts in the month it was resolved (`RESOLVED_DATE`), whatever the reason for it.\nMoney back to the card is money returned to the card or wallet the customer paid the order with.\n\nPublish the customer care data source representing money back to customers for 2024, at a monthly grain, ascending (`refunds_money_back_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the refunds were resolved in, as YYYY-MM. |\n| `refunds` | integer | Refunds resolved in the month, every reason. |\n| `refunded_usd` | number | The amounts those refunds were resolved for, in dollars. |\n| `back_to_card_usd` | number | The part of that which went back to the card or wallet the customer paid with, in dollars. |", "version": "d886dd0c896886eb820517508f0369b59aacccce58bc826d3ed83b0182362c47"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "The month the order was placed, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the month that Busy Kitchen held.", "name": "orders_held", "type": "number"}, {"description": "Minutes the holds added to those orders' quoted prep times, in total.", "name": "hold_minutes", "type": "number"}, {"description": "Storefronts that held at least one order in the month.", "name": "storefronts", "type": "number"}], "description": "For every month of 2024, the orders Busy Kitchen held, the minutes the holds added to quoted prep times, and the storefronts that held orders.", "key": ["month"], "max_rows": 5000, "name": "busy_kitchen_holds_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Merchant operations › Busy Kitchen holds, 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-83-busy-kitchen-holds", "prompt": "Merchant operations needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe Busy Kitchen tile shows how often kitchens were swamped during 2024. A storefront that sets its own prep times can switch on Busy Kitchen: when too many tickets are waiting, new orders are held for a few minutes before they reach the kitchen, and each held order's quoted prep time is pushed out by the time it was held. A big order is quoted a longer prep time too, however quiet the kitchen is; that is not a hold. Rebuild the tile from the warehouse and publish it.\n\nAn order counts in the month it was placed, whatever happened to it afterwards.\n\nPublish the merchant operations data source representing Busy Kitchen holds in each month of 2024, at a monthly grain, ascending by month (`busy_kitchen_holds_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the order was placed, as YYYY-MM. |\n| `orders_held` | number | Orders placed in the month that Busy Kitchen held. |\n| `hold_minutes` | number | Minutes the holds added to those orders' quoted prep times, in total. |\n| `storefronts` | number | Storefronts that held at least one order in the month. |", "version": "d945bbf7bad1961a825dc8c782ac8c93aa5f48923e60942a6572591db340f906"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "The month the order was placed, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the month whose POS told us when the food was ready.", "name": "orders", "type": "number"}, {"description": "Of those, orders whose food was ready at or before the quoted ready time.", "name": "ready_on_time", "type": "number"}, {"description": "Of those, orders whose food was ready after the quoted ready time.", "name": "ran_late", "type": "number"}], "description": "For every month of 2024, the orders whose POS told us when the food was ready, and how many were ready by the quoted ready time or ran late.", "key": ["month"], "max_rows": 5000, "name": "prep_accuracy_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Merchant operations › Prep accuracy against the quote, 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-84-prep-accuracy-pos", "prompt": "Merchant operations needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe prep accuracy tile shows how well kitchens hit the prep times we quoted during 2024. It covers only the orders whose POS told us when the food was ready: for every month, how many of those orders there were, how many had the food ready by the quoted ready time, and how many ran late. Rebuild the tile from the warehouse and publish it.\n\nAn order counts in the month it was placed.\n\nPublish the merchant operations data source representing prep accuracy in each month of 2024, at a monthly grain, ascending by month (`prep_accuracy_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the order was placed, as YYYY-MM. |\n| `orders` | number | Orders placed in the month whose POS told us when the food was ready. |\n| `ready_on_time` | number | Of those, orders whose food was ready at or before the quoted ready time. |\n| `ran_late` | number | Of those, orders whose food was ready after the quoted ready time. |", "version": "87ec7d911838c1aa6d6a19144137738a36eaf42ab8de76985e3d6bc51db35653"} {"area": "Growth, Marketing & Membership", "contracts": {"data_sources": [{"columns": [{"description": "The New York calendar day, YYYY-MM-DD.", "name": "day", "type": "date"}, {"description": "Hours couriers spent on shift during the day, the whole fleet together.", "name": "courier_hours", "type": "number"}, {"description": "Average number of couriers on shift at any given moment of the day.", "name": "avg_couriers_on_shift", "type": "number"}], "description": "For every day of the first quarter of 2024, the hours couriers spent on shift during that New York day.", "key": ["day"], "max_rows": 5000, "name": "courier_shift_hours_daily_q1_2024", "order_by": ["day"], "row_count": 91, "title": "Courier operations › Time on shift by day, Q1 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "clk-01-courier-shift-hours-q1", "prompt": "Courier operations is adding a daily supply line to its quarterly review and needs the data source behind it rebuilt on the new reporting schema: the hours couriers spent on shift on each day of the first quarter of 2024, and how many couriers that meant were on shift at a time.\n\nA courier is on shift from logging on until logging off, whether or not a job comes in. Days are New York calendar days, midnight to midnight, and each day gets only the time that fell inside it: a late shift's hours after midnight belong to the next day.\n\nPublish the courier operations data source for time on shift by day, one row per day from 1 January to 31 March 2024, ascending by day (`courier_shift_hours_daily_q1_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `day` | date | The New York calendar day, YYYY-MM-DD. |\n| `courier_hours` | number | Hours couriers spent on shift during the day, the whole fleet together. |\n| `avg_couriers_on_shift` | number | Average number of couriers on shift at any given moment of the day. |", "version": "5afaf48dbe1795a2b853d46cf462e446f0a526e8f21406dbb4a2f170e6a7254f"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "nec-01-courier-1099-nec", "prompt": "It's mid-January 2025 and Tax is getting the courier 1099-NEC run ready. Every courier we paid\n$600 or more during 2024 gets a form, and the form shows everything we paid them in 2024 (pay,\ntips, incentives and adjustments), before the fee we charge for instant cash-outs. Report how many\ncouriers get a form and the total the forms will show, as metrics couriers_filed and\nreported_compensation_usd. Some couriers have already asked what their form will say: report the\namount for each of these as a metric named by their VENDOR_ID, and 0 for any who won't get one:\n2312623, 2249556, 2318422, 2247983, 2329694, 2328653, 2329356, 2329111, 2314320, 2250948,\n2250197, 2312734, 2298209, 2272463, 2262787, 2287690, 2296620, 2246066, 2301979, 2241642.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "4c40438c1af38c259d1171b7a95e4b8f7452f90e6d5b8847442a86e379c5562a"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-12", "domain": "Customer Analytics", "id": "vd-c1-early-order-repeat-share", "prompt": "Growth thinks customers are forming restaurant habits more slowly than they used to. The team is interested in the early-order repeat share, i.e., of the orders that were a customer's 2nd, 3rd or 4th order on the platform, the share placed with a merchant that customer had already ordered from.\n\nA customer's order number counts every order the customer has placed on the platform since opening their account, cancelled orders included, in the order they were placed. Orders placed in the same second are taken in order-number order. An early order is a repeat if the same customer placed an earlier order (regardless of whether it is cancelled or not) with the same merchant. Months are calendar months in the warehouse's own local time.\n\nCompute it for early orders placed in January 2024 and for early orders placed in December 2024 (unit=\"count\"). Report the following metrics:\n- `early_order_repeat_share_2024_01` (a fraction between 0 and 1)\n- `early_order_repeat_share_2024_12` (a fraction between 0 and 1)\n- `early_orders_2024_01` (how many early orders the January share is taken over)\n- `early_orders_2024_12` (how many early orders the December share is taken over)", "version": "882c04860b0238f4af58072b908b3f9c31b0789e763a2263cfe798eccbf2d14d"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-12", "domain": "Customer Analytics", "id": "vd-c2-six-month-membership-retention", "prompt": "Membership leadership benchmarks the program against a published survivor curve: of people who sign up, 36% are still subscribed six months later. They want the same figure for the customers who became members in January 2024.\n\nA new member is a customer whose first self-paid membership on record started in January 2024. A partner-sponsored comp is not a sign-up and a customer with an earlier self-paid membership on record is not new. A membership that began with a free trial started on the trial's first day and the trial is part of that self-paid membership. A customer who never converted from the trial to a paid plan still joined and left when the trial ended. A membership is in force on a day if it started on or before that day and had not yet ended. It ends on the date it was terminated or, if it was never terminated, on the date its current period ends. Self-paid and sponsored memberships both count as in force.\n\nThe six-month anniversary is the same day of the month six months after the member's start date (a day that does not exist in that month becomes the month's last day). A new member is retained if they were subscribed without interruption from the day they joined through their six-month anniversary.\n\nCalculate the following metrics (unit=\"count\"):\n- `january_2024_new_members`\n- `retained_six_months`\n- `six_month_retention` (retained over new members, a fraction between 0 and 1)", "version": "7f865b4f6f72ccdadc6625d6a83455312e1e0b7267d7faea133482eeb4664213"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-12", "domain": "Customer Analytics", "id": "vd-c3-twelve-month-membership-retention", "prompt": "The board deck needs the membership programme's 12-month retention next to the 6-month figure.\n\nA sign-up is a customer's first self-paid membership on record. Partner-sponsored comps are not sign-ups. A membership that began with a free trial started on the trial's first day and the trial is part of that self-paid membership. A customer who never converted from the trial to a paid plan still joined and left when the trial ended. A membership is in force on a day if it started on or before that day and had not yet ended. It ends on the date it was terminated or, if it was never terminated, on the date its current period ends. Self-paid and sponsored memberships both count as in force.\n\nThe N-month anniversary is the same day of the month N months after the member's start date (a day that does not exist in that month becomes the month's last day). A sign-up is retained at N months if the customer was subscribed without interruption from the day they joined through their N-month anniversary.\n\nMeasure each figure on every sign-up these records let you measure it on, pooled across start months. A group of sign-ups counts only if these records hold everyone who signed up in it, including those who later left. The board wants observed figures.\n\nReport these numbers (unit=\"count\"):\n- `six_month_retention` (a fraction between 0 and 1) and `six_month_cohort` (how many sign-ups it is measured on)\n- `twelve_month_retention` and `twelve_month_cohort`, likewise\n\nIf a figure cannot be measured from these records, report its retention as -1 and its cohort as 0. Then file a note saying which sign-ups each figure covers and why.", "version": "e1f0c4817bec389adb609e096ee157cfc7a06a2869707061e77baedc9f76fd15"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-12", "domain": "Finance & Accounting", "id": "vd-j2-membership-order-economics", "prompt": "Members order far more often than non-members, and someone has proposed that this makes the membership program obviously profitable. Before that argument goes to management, Finance wants the underlying numbers for calendar 2024.\n\nContribution margin (CM) of an order is the platform's own P&L on it and nothing else: plus the merchant's commission; the delivery, service, small-order and regulatory response fees the customer actually paid; and, on a member's order, the part of the member's subscription fee that the books allocate to that order; less the courier's pay for the order, including any minimum-pay true-up the books accrue to that delivery; the platform-funded share of the order's promotion; the platform-funded share of its refunds; card processing; a chargeback the platform lost; compensation paid to the merchant for a cancellation; referral bonuses paid on the order; and courier quest bonuses booked on the order.\n\nTips, sales tax, the merchant-funded shares of promotions and refunds, and a member's pickup credit (a tender, like a gift card) are not part of CM. An order that was cancelled and never charged still has a CM: whatever the platform paid out on it, negative. An order's CM does not depend on whether the journal that carries it has been posted yet.\n\nA member order is any order placed in 2024 while the customer held an active membership, whether or not the storefront was eligible for member benefits. Every other 2024 order is a non-member order. Averages are over all orders in the group, cancelled ones included. Subscription revenue recognised in 2024 is subscription fees recognised as revenue in the 2024 books (not cash billed, i.e., an annual plan is recognised over its term). Count every 2024 journal, posted or not. Platform CM including subscriptions is total CM over every 2024 order plus the subscription revenue recognised in 2024, with every subscription dollar counted once.\n\nCalculate the following values in USD:\n- `cm_per_member_order`\n- `cm_per_nonmember_order`\n- `subscription_revenue_recognised`\n- `platform_cm_including_subscriptions`", "version": "cef1e1541228c7c5848769d8f16aae85ea56b38e6f639156395d380414aee43a"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Finance & Accounting", "id": "vd-j4-negative-cm-share", "prompt": "The board has heard that the platform loses money on a large share of its orders and wants the number, measured properly, for calendar 2024.\n\nContribution margin (CM) of an order is the platform's own P&L on it and nothing else: plus the merchant's commission; the delivery, service, small-order and regulatory response fees the customer actually paid; and, on a member's order, the part of the member's subscription fee that the books allocate to that order; less the courier's pay for the order, including any minimum-pay true-up the books accrue to that delivery; the platform-funded share of the order's promotion; the platform-funded share of its refunds; card processing; a chargeback the platform lost; compensation paid to the merchant for a cancellation; referral bonuses paid on the order; and courier quest bonuses booked on the order.\n\nTips, sales tax, the merchant-funded shares of promotions and refunds, and a member's pickup credit (a tender, like a gift card) are not part of CM. An order that was cancelled and never charged still has a CM. Whatever the platform paid out on it, negative. An order's CM does not depend on whether the journal that carries it has been posted yet.\n\nThe population is charged orders, which are defined as orders placed in 2024 for which the customer was actually invoiced (a cancelled order that was never charged is not in it, whereas a cancelled order the customer was billed for is). An order loses money when its CM is -$0.01 or less. A member order is any order placed while the customer held an active membership, whether or not the storefront was eligible for member benefits.\n\nCalculate the following metrics:\n- `charged_orders` (unit=\"count\")\n- `charged_orders_losing_money` (unit=\"count\")\n- `share_losing_money` (unit=\"ratio\", a fraction between 0 and 1, of charged orders)\n- `share_losing_money_member_orders` (unit=\"ratio\", of charged member orders)\n- `share_losing_money_nonmember_orders` (unit=\"ratio\", of charged non-member orders)", "version": "e3bd27745aa0e0f5ebf40b660a4044b69dae0e84d31fdc696baa681cc64dc279"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Operations Research", "id": "vd-f2-quotes-priced-off-another-storefront", "prompt": "Kitchen Ops is auditing the platform's prep-time quotes for calendar 2024.\n\nWhen the platform, rather than the merchant, sets an order's prep quote at confirmation, it starts from the running prep estimate it keeps for a storefront (every update to that estimate is logged) and adds a load allowance of 0.75 minutes for each ticket already open in that kitchen, rounded to a tenth of a minute. The open-ticket count is not published, so a quote is consistent with an estimate when the quote minus the estimate in force at the quote instant equals 0.75 x k for some whole k >= 0, within 0.05 minutes.\n\nKitchen Ops suspects that at some storefronts, platform-set quotes were priced from a **different storefront's** estimate. Find every storefront where at least one 2024 platform-set quote was priced from another storefront's estimate, and name that other storefront.\n\nFile **one** `mission_control.report_metrics` call with `unit=\"count\"` and one item per such storefront:\n\n* `name`: the storefront's supplier id (`VENDOR_ID`), as text\n* `value`: the `VENDOR_ID` of the storefront whose estimate priced those quotes\n\nIf there are none, file nothing. Do not report a storefront you have not traced to rows.", "version": "78342d4eba3410a43da9aa8344f65b59dc14ffc44b8906be4169fc5f1dc9807b"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Compliance", "id": "vd-e1-dcwp-consumer-fees-1", "prompt": "NYC's Department of Consumer and Worker Protection (DCWP) publishes the average consumer fees per delivery in a quarter for each app. Policy wants our Q4 2024 figure computed on DCWP's own basis before it is compared with the published panel.\n\nConsumer fees include per-delivery charges as well as charges for subscription services. Every fee the customer is charged on a delivery order counts (delivery, service, small-order and regulatory fees). However, the food subtotal, sales tax and tips do not count.\n\nFees are what consumers actually paid. Where a membership benefit waived or reduced a fee, the fee counts at what the member was actually charged, not at its list price. Promotional coupons are treated as a discount on the order, not as a reduction of fees, so they do not change this figure. Subscription charges are the membership charges members actually paid during the quarter, each counted in full in the quarter it was paid, at the amount charged. Per delivery means divided by completed deliveries: delivery orders (not pickup) placed in the quarter that were not cancelled. Order fees are those charged on these same orders.\n\nQ4 2024 is 1 October to 31 December 2024, New York time.\n\nDetermine the following values (unit=\"usd\"):\n- `consumer_fees_per_delivery` (the DCWP figure per delivery)\n- `order_fees_per_delivery` (the per-delivery charges alone per delivery)\n- `subscription_charges_per_delivery` (the subscription charges alone per delivery)", "version": "964ce81e142f92781fc64fea8b1a68ab13c6c8ba27e0d290ce8d152a13c23671"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Finance & Accounting", "id": "fin-01-collections-past-due", "prompt": "We're getting ready to send payment reminders in January 2025. Work out which customer invoices still open in the extract were more than 90 days overdue on December 31, 2024, and how much each customer still owes us.\nSome invoices look unpaid even though we've already received the money. Before collecting, offset any unapplied cash on receipts linked to that invoice. Leave out reversed receipts and cash from unrelated receipts. Only pursue invoices with more than $5.00 left to collect. Leave membership subscription invoices out of every figure and the filing; Subscriptions runs its own dunning for those.\n\nReport through **one** `mission_control.report_metrics` call, using these names exactly:\n\n* `open_invoices_count` (`unit=\"count\"`) — open customer invoices at extract, all ages\n* `open_invoices_usd` — their total remaining balance\n* `unapplied_cash_usd` — cash received from customers that is not applied to any invoice, across all receipts\n* `collectible_invoices_count` (`unit=\"count\"`) — invoices on Collections' list (open and more than 90 days past due at December 31) with more than $5.00 genuinely collectible once unapplied cash on their own receipts is taken into account\n* `collectible_usd` — the total genuinely collectible amount on those invoices\n\nThen file the collection attempts in **one** call:\n\n```\nmission_control.remediate_payments(\n items=[{\"order_id\": , \"amount\": }, ...],\n action=\"recapture\", reason=Reason.OTHER)\n```\n\n— one item per collectible invoice, keyed by the **order** the invoice bills (Order Management's `HEADER_ID`), `amount` the collectible part of that invoice.", "version": "e658eb7f40a0fcdb1ea7ad31714e9de9a96db256578e2a09226c79df9f749949"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Finance & Accounting", "id": "fin-07-ar-to-gl-reconciliation", "prompt": "We need to tie out customer receivables at December 31, 2024. Reconcile general ledger account 1101 to the Receivables subledger and explain the difference.\n\nDecember is still open, so include every journal effective on or before December 31, whether it's posted or unposted. For the subledger, use invoices dated on or before December 31. The catch is that we only have today's open-item file, which already includes January activity. You'll need to account for January settlements as well as anything in the ledger that hasn't made it into the subledger.\n\nFile the reconciliation as one schedule through **one** `mission_control.file_schedules([...])` call: `name=\"ar_to_gl_reconciliation\"`, `as_of=\"2024-12-31\"`, `key_columns=[\"item\"]`, one column `amount` (ledger sign convention throughout: debits positive). The first two rows are the two positions; every further row is a reconciling item, signed so that the items sum to the difference between the two positions. Use these labels for the `item` column and no others:\n\n* `gl_receivable` — the ledger balance of the receivable at December 31\n* `ar_invoices_remaining` — the remaining balance on those invoices in the subledger today\n* `applied_cash_not_yet_settled` — cash the subledger has applied to invoices whose settlement had not been advised by December 31\n* `returned_receipts_not_yet_clawed_back` — payments settled by December 31 and returned after it: the subledger has reopened the invoice, the ledger has not yet recorded the return\n* `refund_credit_memos` — refund credit memos the subledger carries against those invoices\n* `approved_adjustments` — approved subledger adjustments (credits and write-offs) against those invoices\n* `customer_credits_tendered` — account credit the subledger applied against those invoices\n* `on_account_credit_balances` — credit balances the subledger carries on account for customers\n* `invoices_not_recognized_in_gl` — the negative of invoice amounts included in the subledger population but not yet recognized in account 1101 by year-end; establish the recognition event from the accounting records\n* `incomplete_invoices_gl` — the ledger balance carried by invoices the autoinvoice run left incomplete (they have no open item in the subledger)\n\nNot every label is a reconciling item in this ledger. File a row only for an item that actually explains part of the difference; a label whose amount is zero, or that describes a balance the receivable account never carried, must not be filed — a row that does not belong is as wrong as a missing one. State the reconciliation in a `mission_control.note(...)`: ledger receivable less the subledger's invoice balance must equal the sum of the reconciling items you filed, to the cent — and say which label, if any, you deliberately left out and why.", "version": "1df242430df57124ea3ab745897053870a2b5db8b8eb2253201e89d665669290"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Finance & Accounting", "id": "fin-15-customer-credit-liabilities", "prompt": "We need to check how much we owed customers at the end of 2024. Reconcile customer refund payable, account 2105, and customer credit payable, account 2107, to the Receivables subledger as of December 31.\nDecember is still open, so include every journal effective on or before December 31, posted or unposted. The subledger shows today's balances, including January credit memos, receipts and applications, and some December items have since settled. Work back to what was owed at year-end.\nAccount credit is recorded as a credit memo on account. A card refund is a credit memo settled by a negative receipt. One credit memo can cover more than one natural account, so work out how much belongs to each liability.\n\nFile one schedule through **one** `mission_control.file_schedules([...])` call: `name=\"customer_credit_liabilities\"`, `as_of=\"2024-12-31\"`, `key_columns=[\"item\"]`, one column `amount` (ledger sign: a credit balance is negative). Use these labels and no others:\n\n* `gl_refund_payable` — the 2105 balance at December 31\n* `gl_credit_payable` — the 2107 balance at December 31\n* `card_refunds_in_flight` — card refunds resolved by December 31 whose settlement had not been advised by then\n* `platform_credits_cash_share` — account credits granted in lieu of a card refund, at the amount the refund payable still carries for them\n* `referral_credits_unspent` — referral credit granted by December 31 and not yet spent by then\n* `refund_credit_remainders_unspent` — the share of refunds that was returned as spendable account credit and not yet spent by December 31\n* `on_account_credit_balances` — the credit balances the subledger carries on account today\n* `refund_credit_memos_open` — refund credit memos still open in the subledger today\n* `card_refunds_settled` — card refunds settled through the processor by December 31\n* `chargebacks_in_flight` — lost disputes not yet settled by December 31\n\nNot every label is part of either proof: file a row only for an item that is a component of one of the two balances — a row that does not belong is as wrong as a missing one. State both proofs in a `mission_control.note(...)`: each ledger balance must equal the sum of its components to the cent, and say which labels you left out and why.", "version": "47efd844b62681bdede2893415bb3a5b96c572803e41058d605c4adced91dec5"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-12", "domain": "Finance & Accounting", "id": "mer-32-compensation-over-commission", "prompt": "Are we paying any storefronts more in cancellation compensation than we're earning from them in commission? Find every storefront where that happened in 2024. The compensation is what we pay when a customer cancels during prep and we waive the cancellation charge. Compare it with the storefront's total commission across all its charged orders in 2024.\n\nFile one `compensation_over_commission` schedule using `mission_control.file_schedules([...])`, with `key_columns=[\"vendor_id\"]` and columns `vendor_id`, `compensation_usd` and `commission_usd`. Give amounts to the cent and include one row per storefront whose compensation exceeded its commission.", "version": "6982d27d39112174890e157e29572d7577c9682fe39af5bee288c096d35e4951"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "rf-01-platform-refund-spend-december", "prompt": "It's November 30, 2024, and we're working out how much to set aside for December customer refunds. Forecast how much of the refunds resolved in December we'll have to fund ourselves.\n\nUse `XX_REFUNDS.RESOLVED_DATE` to put refunds in the right month and add up `PLATFORM_FUNDED_AMOUNT`. Count our share only, regardless of when the order was placed or the refund requested. Treat dates as New York local time.\n\nUse only information available through November 30; don't query December or later data. File one `platform_funded_refunds_usd` forecast, a point estimate and an 80% central prediction interval.", "version": "261c6bfc102429d10d1063b325e69b307239a269fb842da8d1efab6734172a57"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Dashboards", "id": "dl-01-december-late-delivery-rate", "prompt": "We're planning December delivery capacity. What percentage of our deliveries do you expect to arrive late in December 2024?\n\nFor each month, use delivery legs with a non-null `XX_DELIVERY_LEGS.DELIVERED_DATE` in that month and a matching order promise through `XX_ORDER_PROMISES.HEADER_ID`. A leg is late if `DELIVERED_DATE` is strictly after `PROMISED_DELIVERY_AT`. Count each leg once and divide late legs by all matched delivered legs. Report a percentage, so 15.0 means 15%. Treat dates as New York local time.\n\nUse only information available through November 30; don't query December or later data. File one `late_delivery_pct` forecast, a point estimate and an 80% central prediction interval.", "version": "e3e3c4ca6758a0c6a8c04898b0a63797a92058b7789cde4d12cbdd112b0e5b57"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "cb-01-december-lost-disputes", "prompt": "How much do we expect to lose on card disputes that resolve in December 2024? Finance needs a forecast for the month.\n\nAdd up `AMOUNT` for disputes with `OUTCOME_CODE='LOST'`, assigning them to months by `XX_DISPUTES.RESOLVED_DATE`. Include disputes over earlier orders or opened in earlier months if they resolve in December. Leave out disputes we win, and treat dates as New York local time.\n\nUse only information available through November 30; don't query December or later data. File one `lost_disputes_usd` forecast, a point estimate and an 80% central prediction interval.", "version": "beffda94903e4a5b0f3de2b7aa749299860e66a42400455bd213cf591ecd2bc8"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "pe-01-december-posted-promo-expense", "prompt": "Marketing wants to know how much promotion expense we'll book in December 2024. Forecast the amount that will hit the ledger, including credits and reversals.\n\nUse accounted debits less accounted credits on natural account `5201`, PROMO EXPENSE (`GL_CODE_COMBINATIONS.SEGMENT3`). Include detail combinations only (`SUMMARY_FLAG='N'`) and posted journals only (`GL_JE_HEADERS.STATUS='P'`), assigning lines to months by `EFFECTIVE_DATE`. Use that same definition for prior months; the figure we need is booked expense, not the total discounts customers received. Treat dates as New York local time.\n\nUse only information available through November 30; don't query December or later data. File one `posted_promo_expense_usd` forecast, a point estimate and an 80% central prediction interval.", "version": "ade6a6ffb5df615a2741ed4876b319473c5ecc6831a13c5827d46dc353036347"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "mr-01-december-membership-revenue", "prompt": "How much membership revenue do we expect to recognize in December 2024? Finance needs a forecast of the revenue on the books, rather than the cash coming in from renewals.\n\nUse accounted credits less accounted debits on natural account `4103`, MEMBERSHIP REVENUE (`GL_CODE_COMBINATIONS.SEGMENT3`). Include detail combinations only (`SUMMARY_FLAG='N'`) and posted journals only (`GL_JE_HEADERS.STATUS='P'`), assigning lines to months by `EFFECTIVE_DATE`. Use the same definition for prior months and treat dates as New York local time.\n\nFile one forecast named `membership_revenue_usd`, unit `usd`, horizon `2024-12`, with a point estimate and an 80% central prediction interval.", "version": "7cf3e7b2098b381d80ea2ac0ae77631d0dfce9aa975d70e32def8431a15b1e7a"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "pf-01-december-card-processing-expense", "prompt": "We need to budget for card processing in December 2024. Forecast how much card-processing expense we'll book that month.\n\nUse accounted debits less accounted credits on natural account `5101` (`GL_CODE_COMBINATIONS.SEGMENT3`). Include detail combinations only (`SUMMARY_FLAG='N'`) and posted journals only (`GL_JE_HEADERS.STATUS='P'`), assigning lines to months by `EFFECTIVE_DATE`. Use the same definition for prior months and treat dates as New York local time.\n\nFile one forecast named `card_processing_expense_usd`, unit `usd`, horizon `2024-12`, with a point estimate and an 80% central prediction interval.", "version": "ab34c585f6b31b8bed644161025cd77eef16f1e128ea78e1b9b6d6d57e7a7098"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "oc-01-december-order-cancellation-rate", "prompt": "We're planning staffing and merchant support for December 2024. What percentage of orders do you expect to be cancelled?\n\nFor each month, take all orders placed that month using `OE_ORDER_HEADERS_ALL.ORDERED_DATE`, whatever their status. Count an order as cancelled if its header has `CANCELLED_FLAG='Y'`, including customer cancellations and storefront rejections. Divide cancelled orders by all orders placed and report a percentage, so 3.0 means 3%. Treat dates as New York local time.\n\nFile one forecast named `order_cancellation_pct`, unit `pct`, horizon `2024-12`, with a point estimate and an 80% central prediction interval.", "version": "261a62b854f075161f6e76a99a965fdbb99bb3eca802983a91eab7d6b76e6d6c"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "cs-01-december-support-intake", "prompt": "We're putting together the December support rota. How many customer support cases should we expect to be opened in December 2024?\n\nCount each `XX_SUPPORT_CASES.CASE_ID` once, using `OPENED_DATE` to assign it to a month. Include every category and channel, whether the case escalates or not, whenever it closes, and even if the order was placed earlier. Treat dates as New York local time.\n\nFile one forecast named `support_cases_opened`, unit `count`, horizon `2024-12`, with a point estimate and an 80% central prediction interval.", "version": "980605e1cd8b0c75652f5ae3b1c26214a6092a43b6c10134b44c27f264d03221"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "rv-01-december-customer-review-volume", "prompt": "Our marketplace quality team is planning its December staffing. Forecast how many customer reviews we'll receive in December 2024.\n\nCount distinct `XX_REVIEWS.REVIEW_ID` values across all star ratings and storefronts. Use `REVIEW_DATE` to assign each review to a month, rather than the order date or when the record entered the warehouse. Treat dates as New York local time.\n\nFile one forecast named `customer_reviews_submitted`, unit `count`, horizon `2024-12`, with a point estimate and an 80% central prediction interval.", "version": "5e5bba753acfda01ba1cd74b8077488203c7473aa8d9ce341ec8d28c9bccad79"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-18-december-after-confirm-rejections", "prompt": "Sometimes a merchant confirms an order and then rejects it, leaving us to help the customer. We're planning the team that handles these cases. How many of these rejections should we expect in December 2024?\n\nUse the available `food_delivery` warehouse and only information available through November 30. Count `XX_ORDER_REJECTIONS` rows by `REJECTED_DATE` where `REJECT_STAGE_CODE='AFTER_CONFIRM'`. Count each rejection event, so two qualifying events on the same order count twice. When one storefront's kitchen is swamped, the team handles its cases for that day as one batch, so count at most 5 events per storefront (`VENDOR_ID`) per calendar day of `REJECTED_DATE`: a storefront with 40 qualifying events on one day counts 5, one with 3 counts 3. The December figure is the sum of these capped storefront-day counts over the days of December. File one `file_forecasts` item named `december_after_confirm_rejections`, with `horizon='2024-12'`, a count-valued point estimate, and an 80% central prediction interval (`lower`, `upper`, `level=0.8`).", "prompt_edit": "the warehouse's internal dataset name is replaced by its public name, food_delivery", "version": "d747867515925160c7f78b1756e59c7ae31866101e91894864aea1c0be83b255"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-19-december-customer-otp-failures", "prompt": "It is November 30, 2024. Using the food_delivery warehouse available on this date, the customer account team needs a forecast of failed one-time-password challenges in December 2024 for account support staffing. Count `XX_CUSTOMER_ACCOUNT_EVENTS` rows with `EVENT_DATE` in December, `EVENT_TYPE_CODE='OTP_CHALLENGE'`, and `OUTCOME_CODE='FAILED'`. Count challenge event rows, including multiple failed attempts by one account. File one `file_forecasts` item named `december_failed_otp_challenges`, with `horizon='2024-12'`, a count-valued point estimate, and an 80% central prediction interval (`lower`, `upper`, `level=0.8`). Use only information available through November 30 and explain the pattern you used in a note.", "prompt_edit": "the warehouse's internal dataset name is replaced by its public name, food_delivery", "version": "bd5b002a6f43552e1b7ece5e35ec66336d4914573be35751ccd9df5aa174d9b3"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-24-december-pos-prep-activations", "prompt": "It is November 30, 2024. Using the food_delivery warehouse available on this date, merchant integrations needs a forecast of December merchant activations whose prep-time source is POS-integrated. Count `XX_MERCHANT_INTEGRATIONS` rows with `ACTIVATED_DATE` in December 2024 and `PREP_TIME_SOURCE_CODE='POS_INTEGRATED'`. Count integration activation rows, not distinct POS vendors or activations still live at year-end. File one `file_forecasts` item named `december_pos_prep_activations`, with `horizon='2024-12'`, a count-valued point estimate, and an 80% central prediction interval (`lower`, `upper`, `level=0.8`). Use only information available through November 30 and explain your assumptions in a note.", "prompt_edit": "the warehouse's internal dataset name is replaced by its public name, food_delivery", "version": "6c085b2c7e148a6736bb8f953dfaf3a953914b3a20851997c14402f3efdac938"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "gw-13-december-courier-wait-minutes", "prompt": "We're planning December courier coverage. How much time will couriers spend waiting at merchants, how much of that wait will be avoidable, and how many waits will qualify for wait pay?\n\nA wait belongs to its arrival date. Use the recorded avoidable portion of each wait.\n\nForecast `merchant_wait_hours`, `avoidable_wait_hours`, and `wait_pay_eligible_events`, with 80% central prediction intervals.", "version": "f3df3ed606d666e67055a626c806f2f987c9b639c60ffeb1c84597f043047ff1"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-14-december-membership-reactivation", "prompt": "We're planning the membership retention team's December workload. How many membership reactivations should we expect?\n\nCount reactivation events when they occur, including repeated reactivations on the same contract.\n\nForecast `december_membership_reactivations`, with an 80% central prediction interval.", "version": "5d5ad739a5b56bb121f91e0c7a81a982a3c378435ce22e8a155f0176cb336888"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-10-december-risk-board", "prompt": "We're setting December holiday staffing. How many unexpected merchant closures should we expect, and what percentage of payment authorization attempts will be declined?\n\nInclude closures recorded as unplanned, equipment failure, or POS offline, by their start date. Count repeated closures separately. For payment authorizations, include retries in both the declined attempts and the total.\n\nForecast `december_unplanned_merchant_closures` and `december_auth_decline_pct`, with 80% central prediction intervals.", "version": "1abdeabf383ca42bf3c1c599e952b61dcdf4b4e7cdc979e3a2888b73ed65e344"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "gw-15-december-delivery-attempt-failures", "prompt": "We're staffing the team that handles failed deliveries in December. How many attempts will fail because the customer can't be reached, the address can't be found, or access is denied?\n\nCount each attempt on its attempt date, including repeated attempts on the same order.\n\nForecast `customer_unreachable_attempts`, `address_not_found_attempts`, and `access_denied_attempts`, with 80% central prediction intervals.", "version": "c33b714998d90ec58cf0c1a69d92c5cd1b3e650317bdddcf6d60d138ad1e5fa3"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-13-december-promo-rejection-mix", "prompt": "Marketing is planning December support staffing. How many promotion-code attempts will fail because the campaign budget is exhausted, the code is exhausted, or the customer isn't eligible?\n\nCount attempts on their attempt date, including retries.\n\nForecast `budget_exhausted_attempts`, `code_exhausted_attempts`, and `not_eligible_attempts`, with 80% central prediction intervals.", "version": "97323639ec619169709593719a680044629e208436547e3ae8ac3afee8392e75"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "gw-21-december-courier-identity-review", "prompt": "We're planning December staffing for courier identity reviews. How many checks will fail, how many of those failures will be confirmed identity mismatches, and how many checks will expire?\n\nA check belongs to its request date. Confirmed mismatches also count among failed checks.\n\nForecast `failed_checks`, `confirmed_mismatches`, and `expired_checks`, with 80% central prediction intervals.", "version": "1337f71bfebae54a681f8ac14fe7b2e5cbc93fc7c48cba35259bcc69aa039e54"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-16-december-holiday-closures", "prompt": "We're planning merchant coverage for December. How many holiday closures should we expect?\n\nCount closure events by their start date, including separate closures at the same merchant.\n\nForecast `december_holiday_closures`, with an 80% central prediction interval.", "version": "15ebdb236fcd2954276b5496b1dbec4bac50f5457dc71db834e80a6fc96c2dc1"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "gw-20-december-customer-account-recovery", "prompt": "We're staffing the December account recovery team. How many account locks, account restorations, and password resets should we expect?\n\nCount completed events on their event date, including repeated events for the same account.\n\nForecast `locked_events`, `restored_events`, and `password_resets`, with 80% central prediction intervals.", "version": "53df3729256b09cc67b43a143b9147328d498790b211976a4e8cea388e5c66fd"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-17-december-campaign-expirations", "prompt": "We're planning the campaign team's December renewal workload. How many campaign authority expirations should we expect?\n\nCount expiration events by their effective date, including repeated events for the same campaign.\n\nForecast `december_campaign_expirations`, with an 80% central prediction interval.", "version": "b5f35dd2d730c36b071776da18c88b975f1de884f4d4cb9153786ac329273b7b"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "gw-24-december-customer-session-dropoff", "prompt": "We're planning December investigations into short customer sessions. How many will occur at home, while roaming, and at the office?\n\nA short session lasts 0–59 seconds, inclusive, and belongs to its start date. Exclude records with negative durations.\n\nForecast `home_short_sessions`, `roam_short_sessions`, and `office_short_sessions`, with 80% central prediction intervals.", "version": "370977792d67118b81e9a64f033410002db9bc714d6fe1a3a649b1e89adac5f8"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "gw-25-december-tip-adjustment-load", "prompt": "We're planning December tip adjustment work. How many tips will be reduced or increased, and how many dollars will the reductions remove?\n\nCount each adjustment on its adjustment date. Add up the reductions as a positive dollar amount, without offsetting increases.\n\nForecast `downward_tip_adjustments`, `tips_removed_dollars`, and `upward_tip_adjustments`, with 80% central prediction intervals.", "version": "2fe5e57eff3fba313459e2109ffdc1e1a6e529790b8ab099343e6d60c138b1df"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-12-december-uncleared-ap-forecast", "prompt": "We're preparing for the December close. How many December-dated AP checks will still be negotiable in the December 31 extract, and how much will they total?\n\nA check belongs to its check date. Count checks that have not yet cleared at the close.\n\nForecast `december_negotiable_checks_count` and `december_negotiable_checks_usd`, with 80% central prediction intervals.", "version": "90506452f7fe46e5b7bda502164a36566fe715e2ab0cd1b0aec8dc75b9254768"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "indigo-24-delivery-lateness-tail-forecast", "prompt": "We're planning December delivery operations. How many orders will miss their delivery promise, and how late will the worst tenth of late orders be?\n\nAn order belongs to its order date. For late orders, measure elapsed minutes past the promised delivery datetime.\n\nForecast `late_deliveries_count` and `p90_late_minutes` (to two decimals), with 80% central prediction intervals.", "version": "0334b0d175bbe37c4f38c3e8be986a593a9eecfa9609918f5a3329bafd998397"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Compliance", "id": "vd-e3-pay-per-connected-hour", "prompt": "Courier Ops noticed that couriers' base pay plus incentives per delivery **fell** from Q1 to Q2 2024, even though NYC's minimum pay rate for app delivery workers rose from $17.96 to $19.56 an hour on 1 April 2024. Before this reaches the press office, restate courier pay for both quarters on the basis NYC's minimum pay rule uses.\n\nThe rule (6 RCNY 7-810), in plain words:\n\n* The minimum is a rate **per hour of connected time**: all the time a courier is on shift and available for work, whether on a trip or waiting for one. It is not a rate per delivery.\n* The pay that counts toward it is everything the platform pays the courier for the work, including incentives and any amount the platform adds to bring couriers up to the minimum. Customer tips never count.\n* Pay is settled in weekly pay periods, Monday to Sunday, New York time.\n\nFor this restatement:\n\n* **Q1** is the 13 pay weeks from 1 January to 31 March 2024; **Q2** is the 13 pay weeks from 1 April to 30 June 2024.\n* A quarter's **pay** is what the weekly pay statements for its pay weeks credited to couriers, tips excluded. An instant cash-out only lets a courier withdraw pay early: it is not additional pay, and what it cost the courier is already reflected on the weekly statement.\n* **Connected time** is each courier's shift as the shift log records it, from shift start to shift end. A shift that runs across the boundary between quarters counts in each quarter for the part inside it.\n* **Deliveries** are delivery legs completed within the quarter.\n\nFile:\n\n1. **One** `mission_control.report_metrics` call, `unit=\"usd\"`, with these names exactly: `courier_pay_per_delivery_q1`, `courier_pay_per_delivery_q2`, `courier_pay_per_connected_hour_q1`, `courier_pay_per_connected_hour_q2`.\n2. **One** `mission_control.report_metrics` call, `unit=\"ratio\"`, with `deliveries_per_connected_hour_q1` and `deliveries_per_connected_hour_q2`, to three decimals.\n\nDo not report a number you have not traced to rows.", "version": "2d2a92896dd9389cdfe35cd58a235d0b0b607e31565b24fa219d0946fd621bc2"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Compliance", "id": "vd-e6-consumer-fee-peak-quarter", "prompt": "Finance is preparing the quarterly comparison against NYC DCWP's delivery-app reports and wants to know which quarter of 2024 had the highest **consumer fees per delivery**.\n\nUse DCWP's basis:\n\n* **Consumer fees** are every dollar customers paid the platform, other than the price of the food, sales tax and tips.\n* Per-order charges count at what the customer was actually charged on the order, in the quarter the order was placed, and only on orders that are **deliveries**: a delivery order (not a pickup) that was not cancelled. Ignore coupons and any later refunds.\n* Any other payment counts in the quarter it was collected.\n* **Per delivery** means the quarter's consumer fees divided by the quarter's deliveries.\n\nFile **one** `mission_control.report_metrics` call with `unit=\"usd\"` and these names exactly, each the quarter's consumer fees per delivery:\n\n* `consumer_fees_per_delivery_2024q1`\n* `consumer_fees_per_delivery_2024q2`\n* `consumer_fees_per_delivery_2024q3`\n* `consumer_fees_per_delivery_2024q4`\n\nand **one** `mission_control.report_metrics` call with `unit=\"count\"` and the single name `highest_consumer_fee_quarter`, valued 1, 2, 3 or 4.\n\nDo not report a number you have not traced to rows.", "version": "8f6286bcdf0a764f6d20af1cb4d7828c9eea167e71a796ca37eefea4a8657e6c"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "lic-01-borrowed-licences", "prompt": "Compliance thinks some of our storefronts are trading on a permit that belongs to a different business. Work out which ones and take them off the platform. Ban merchants with a call carrying the VENDOR_ID of every storefront you are deactivating.", "version": "e87b32d8d6a01b0f5cd222a2f354f1606fa70998cab8327a34a73f579a5837c1"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "pro-01-campaign-review", "prompt": "Two of our acquisition offers -- promo codes 7812116 and 7842033 seem to be getting unnaturally high levels of usage, and marketing wants to know whether to pull them. Work out who is abusing them and decide what to do about it. We can approach this by banning customers, and/or by ending the promotion entirely. Ban customer accounts using the relevant CUST_ACCOUNT_ID, and/or if you conclude an offer should be pulled, discontinue promo codes with the relevant PROMO_CODE_ID.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "c61ca6bc86bdecce4005bd4bffeafc0fc5d366e4b780f2e420e9c09f0b8195ae"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "ren-01-accounts-worked-by-a-deactivated-courier", "prompt": "Sometimes, after a courier account is banned, they return. This is called re-entry. Our duplicate screen catches the ones who try to register again, so the ones who get through are working an account that belongs to somebody else. Identify those courier accounts. Ban couriers, listing their relevant VENDOR_ID for each account you are deactivating. If you conclude any customer accounts are involved, file those separately with their CUST_ACCOUNT_ID.", "version": "925ee3e3d578d38014c2cfadc44993f7574b2b1d3ff73a7f2535249133635eb5"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "rfd-01-over-reported-orders", "prompt": "To ensure customer satisfaction, we offer the ability for customers to report unsatisfactory orders, such as orders with missing items or that are entirely stolen or lost. These customers are given a refund or store credit. However, it seems like some customers are taking advantage of our policy to over-report unsatisfactory orders. Identify customers that seem to be engaging in this behaviour, and ban them using their CUST_ACCOUNT_ID. If you find that couriers are responsible for these orders instead, file those separately with a call carrying their VENDOR_ID.", "version": "602e45f622ae1996e40109be109639a8cc18e782f4d03442f9d5cd44de150ca6"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "crd-01-fraudulent-payments", "prompt": "We’re concerned about stolen and fraudulent purchases made through our platform. This can be detected by high volumes of payments on a single card, unusual usage patterns associated with a card, or customers with unusual payment patterns. Flag the customers who are likely to be engaging in fraudulent payment behaviour, and ban them using their CUST_ACCOUNT_ID. If you conclude any storefront should also come off the platform, file those bans separately with its VENDOR_ID.", "version": "8c1b203331f566663252bc24ccb77ce24b51648652f8341097807765bb26c979"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-12", "domain": "Fraud", "id": "brs-01-bought-reviews", "prompt": "Some restaurants on the platform benefit from fake reviews to inflate their rankings. This can sometimes be identified by the creation of many new customer accounts, customer accounts that leave many reviews in quick succession, or other unusual customer account behaviour. Identify and ban customer accounts that might be involved in this behaviour, referencing the affected CUST_ACCOUNT_ID.", "version": "667b361ccb3357e03e0b6a734adcaa6641f1bcaed1b724fee998dbb047fd0043"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Fraud", "id": "cnc-01-cancellation-farming", "prompt": "Sometimes, customers, couriers, or businesses need to cancel an order and/or delivery. Stores might cancel an order because they’ve run out of an item, can’t accomodate special requests, or don’t have the capacity to complete the order. Couriers may cancel because they’re unable to find the location of the store or customer. These cancellations might be for benign reasons, but they could also reveal a pattern of fraud or dishonest behaviour. Identify customers, couriers, or businesses that engage in suspicious cancellation activity, and ban them using their VENDOR_ID or CUST_ACCOUNT_ID.", "version": "a4aa8a4c447547426bc0cf4d946a4c4fee57e941da2a1c8a078f34dc90714d48"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "cf-01-expense-variance", "prompt": "Find the five largest (highest net debit) expense accounts of 2024. For each of these accounts,\nperform a Y/Y comparison for the last three months of 2024, comparing each month of 2024 against the same month of 2023.\nFile a schedule called `expense_variance_yoy`, keyed on\n`account` and `period.` This should have columns `current_year`, `prior_year`, and `variance` (as the difference between the two).\nMake sure to format as follows: `account` is the natural account on its own (not the whole code combination) and `period` is `2024-10`, `2024-11`, `2024-12`.", "version": "c26e4d870e14065ac43f0342be4e43acd73914e7958fed0aeb7e34ea060cdf38"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "cf-02-balance-sheet-reconciliation", "prompt": "We're interested in reconciling the balances table. The table contains both summary and base members, and finance wants to ensure that the summary accounts are the sum of the base accounts both for YTD and monthly movement. Reconcile each period of 2024, filing a schedule called `summary_to_detail`, keyed on `period` (using the GL period name, e.g.\n`JAN-24`), with columns `summary_net`, `detail_net` and `variance`. Also report\n`summary_rows_checked` and `mismatched_periods`.", "version": "40a3898166841d6bfda3f8a062aaf9bba22dbd8cd6dd2931b2f27d37e3dd5ad3"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "cf-03-chart-of-accounts", "prompt": "How deep does our chart of accounts actually go? We're interested in finding the deepest member in the chart, `GL_ACCOUNT_HIERARCHIES` structure. We define deepest as the ones with the most levels above them.\n\nReport `max_levels_above` and `n_deepest_accounts`, and file a schedule called\n`deepest_accounts` keyed on `account` listing every account at that depth.", "version": "871a37de48326f917b91dcd05e122d015b3302da99495aca4d765206cc3f6670"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "cf-05-quest-reconcile", "prompt": "Courier Services thinks the quest expense on the books does not match what\nwe actually paid couriers in bonuses in 2024. Reconcile the two and find differences or discrepancies.\n\nFile a schedule called `quest_expense_reconciliation` with the `item` key, with one numeric\ncolumn `amount`, and these five rows: `gl_posted`, `gl_unposted`, `gl_total`,\n`ops_bonus_paid` and `unexplained_variance`.", "version": "832c5e880ac4759dfaabeb1c57387812a7ac0d05dc020d09e362ed789a6fe3ef"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "cf-12-contract-lifecycle", "prompt": "Validate the contract lifecycle for the subscriptions that started in 2024:\nthe status each contract is in should agree with the dates recorded against it.\n\nFile a schedule called `lifecycle_exceptions`, keyed on `check`, with one numeric column\n`n_contracts`. We want to see these checks: `entered_with_date_signed`,\n`terminated_without_date_terminated`, `not_terminated_with_date_terminated`,\n`date_signed_before_start_date`, `date_terminated_after_end_date`,\n`end_date_before_start_date`. Also, report `contracts_in_scope`.", "version": "90532fd95baea47b46a2a7940a4234fa9236f98083546d3c91ca0fd772defbef"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "cf-08-promo-acquisition", "prompt": "We want to check if promo spend actually bring in new customers. For each month\nof 2024, compare what we spent on promos with how many new customer accounts were opened. Then calculate the spend per new account.\n\nFile a schedule called `promo_per_new_account`, keyed on `period` (`2024-01` to `2024-12`). We want to see columns `promo_expense`, `new_accounts` and `promo_per_new_account`.", "version": "2389b8435026241473b8b9ae054f9f27c3e0045faee7f1e622f96e97c06ddf15"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-12", "domain": "Core Finance", "id": "jm-tips-passthrough-jun24-pay4", "prompt": "Legal needs the June 2024 tips certification before we close the month. Tie out the courier tip pass-through and check whether every courier was paid what they are owed under Courier Pay Policy PAY-4 (excerpt below). If anyone was shorted, remediate them.\n\n> PAY-4. Tips on deliveries are the courier's money, never platform revenue; the platform holds them until the courier's weekly payout. The amount collected for a tip is the customer's final tip, collected on delivery. The courier is owed the customer's final tip: a post-delivery tip change replaces the original. Tips on pickup orders belong to the merchant. A tip is paid when the weekly payout for the week its delivery falls in pays out.\n\nReport the following metrics to mission control:\n\n- tips_collected_usd: tips collected on courier deliveries completed in June 2024\n- courier_tips_paid_usd: courier tips paid out on payout dates in June 2024\n- tip_liability_jun30_usd: tips on deliveries completed by 30 June 2024 not yet paid out at 30 June\n- shortfall_orders: June 2024 deliveries whose courier was paid less than they were owed\n- shortfall_usd: the total amount by which those deliveries were short\n\nApply the payment remediation mechanism for every order ID in shortfall with the amount owed. File only what is owed.", "version": "0b51d39a804b09b1317fd3c1edc62a084f6eb23573f556c87c0a274684a94dbc"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-09", "domain": "Forecasting", "id": "wx-01-wet-days", "prompt": "Ops wants to know how much of last quarter's service issues was attributable to weather.\n\nNew York City's daily rainfall is banded the way the National Weather Service bands it, with categories: dry, light, moderate, or heavy. This is based on Central Park's 24-hour total, with thresholds at\n0.2 mm, 2.5 mm and 12.7 mm.\nFor each day in the quarter, we want to predict what the level of rainfall was based on marketplace data alone.\n\nFile a schedule called `wet_days`, with a key `day`. Add a row for each day in the quarter (YYYY-MM-DD), with an entry for `precip_band` which is 'dry,' 'light,' 'moderate,' or 'heavy.' ", "version": "cea92891fd40a5b220e4d0237f4b31b0b99401fc0bb90561911d93520db133c9"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-03", "domain": "Compliance", "id": "rem-01-uncompensated-failures", "prompt": "Not all deliveries are completed successfully. \nThis may be attributed to a number of reasons - the restaurant may be unable to fulfill the order, the courier may lose/steal the order, or the customer may have input incorrect delivery/contact information. \nInvestigate all cases of orders that were not successfully completed. Of those orders, determine which ones were not the fault of the end customer, and proactively offer them a refund - remediate payments with the ORDER_ID specified, carrying the amount we charged them.\nSupport's rule for fault: the order was not the customer's when they gave up after the time the kitchen had quoted their food would be ready, because by then we had already broken a promise to them. A customer who cancelled while the order was still inside that window is owed nothing, and there is nothing to refund where we never took the money.", "version": "f20e073732b45de341f46479ddaad272628be388c0ad25c65f2236193bccf152"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-01", "domain": "Operations Research", "id": "or-04b-coverage-add-january-bare", "prompt": "Operations could have had 15% more online courier-hours in January: 429,000 on top of the 2,861,000 the hourly supply log counts. Where would they have stopped orders being cancelled for want of a courier?", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "623386b6edbba31df61b4099e5dcf10d63bd0819c8f2eff76ad15e0f16a8ee7d"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "The month the order was placed, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the month that were turned away because the kitchen was too full to take them.", "name": "orders_turned_away", "type": "number"}, {"description": "Storefronts that turned away at least one order for that reason in the month.", "name": "storefronts", "type": "number"}], "description": "For every month of 2024, the orders the app turned away because the kitchen was too full to take them, and the storefronts that turned any away.", "key": ["month"], "max_rows": 5000, "name": "kitchen_full_turnaways_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Merchant operations › Kitchen-full turn-aways, 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "sig-01-kitchen-full-turnaways", "prompt": "Merchant operations needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe kitchen capacity tile shows how often restaurants were too busy to take an order during 2024. When a kitchen is too full to get a new order out in time, the app turns the order away itself and tells the customer the restaurant is too busy to take it. That is not the same thing as a restaurant rejecting an order from its tablet, where restaurants pick whatever reason they like. Rebuild the tile from the warehouse and publish it.\n\nAn order counts in the month it was placed.\n\nPublish the merchant operations data source representing orders turned away by full kitchens in each month of 2024, at a monthly grain, ascending by month (`kitchen_full_turnaways_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the order was placed, as YYYY-MM. |\n| `orders_turned_away` | number | Orders placed in the month that were turned away because the kitchen was too full to take them. |\n| `storefronts` | number | Storefronts that turned away at least one order for that reason in the month. |", "version": "f163bbaf166c20d8e0d753fc5785642d3fa2e65e477c959720b8ba6102b11b22"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "col-21-phantom-claimants", "prompt": "Following the missing-item review: the storefronts that were harvesting platform-funded missing-item refunds did it with a small group of regular customers who\nwere in on it. Ban the customer account IDs for those accounts for refund collusion.\n\nThe population you are looking at is every account with two or more missing-item refunds in 2024.\nMost of them are ordinary: people who got a short bag twice, and a minority of habitual claimers who hit many different storefronts.\nClosing a ring member stops the leak, closing anyone else destroys the margin we earn from them, and every decision costs a fixed review amount.", "version": "fea357f926d406a34cbf76287cdc923a24afb91a1cb8c1424b1168105e257103"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "rng-62-reentry-reoffenders", "prompt": "It is the end of 2024, and Trust & Safety is closing the year's appeals ledger.\nThis year we deactivated several thousand courier accounts and let close to two thousand of them back in, most on appeal.\nOnboarding is not the leak: every attempt by a deactivated courier to register again was caught by the duplicate screen or died in background checks.\nThe leak is the appeal. Some of the couriers we reinstated went straight back to what got them noticed: stealing orders, spoofing their location,\nrunning a grab bot, or handing the account to somebody else to work.\n\nFind the reinstated couriers who are at it again and deactivate their vendor IDs, with the reason that fits what you found.\n\nScope: every deactivated courier who was reinstated and is still active / has delivered since the reinstatement.\nCouriers we have already deactivated a second time are out of scope.\nLeaving a re-offender on the road keeps costing us never-arrived refunds and cancellations; deactivating a courier who came back and has behaved destroys what the fleet earns from them, and every decision costs a fixed review amount.\nMost of the people we reinstated were right to appeal, and a bad week after a reinstatement is not a relapse.", "version": "28269d9ab51a3ecf3ea1281b644bc7e598bd292de15e6c57ea2b829188a740c7"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "rng-62-reentry-reoffenders-hard", "prompt": "It is the end of 2024, and Trust & Safety is closing the year's appeals ledger.\nThis year we deactivated several thousand courier accounts and let close to two thousand of them back in, most on appeal.\nOnboarding is not the leak: every attempt by a deactivated courier to register again was caught by the duplicate screen or died in background checks.\nThe leak is the appeal. Some of the couriers we reinstated went straight back to what got them noticed.\n\nFind the reinstated couriers who are at it again and deactivate them. Couriers we have already deactivated a second time are out of scope.", "version": "348fd8b5a1e2d4f829a770f214a7ebc071d56a229855b0037b350c6626ca0382"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-10-shell-storefronts", "prompt": "Growth flagged something in the pickup numbers: a slice of the platform-funded pickup promotions redeemed this year went to storefronts where the customers, the five-star reviews\nand the \"kitchen\" all seem to be the same few people.\nThe pattern they described is a shell: a storefront that exists mostly to sell to its own operator's accounts through pickup, so no courier is ever involved, farming whatever the platform funds: our standing discount every week on a rotating set of accounts, the compensation we pay a storefront when a customer cancels during prep and support waives the charge, and a wall of five-star reviews.\n\nFind the storefronts run as shells and ban their merchant ID for self-dealing.\n\nRemoving a shell stops what it takes from us; removing an honest storefront destroys the margin we earn from it; every review costs a fixed amount.\nBe careful with the small places: a family-run storefront whose relatives and staff order pickup all week on the storefront's own tile promotion and leave five stars looks the same on every ranking\nyou would sort by, and costs us nothing.", "version": "2655f9ae15bc691425881c472d1f365ee498e029d363ad16a5423552c5620373"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "mer-10-shell-storefronts-hard", "prompt": "Growth flagged something in the pickup numbers: a slice of the platform-funded pickup promotions redeemed this year went to storefronts where the customers, the five-star reviews\nand the \"kitchen\" all seem to be the same few people.\nThe pattern they described is a shell: a storefront that exists mostly to sell to its own operator's accounts through pickup, so no courier is ever involved, farming whatever the platform funds: our standing discount every week on a rotating set of accounts, the compensation we pay a storefront when a customer cancels during prep and support waives the charge, and a wall of five-star reviews.\n\nFind the storefronts run as shells and ban their merchant ID for self-dealing.", "version": "efd3422e82b971b93e0a5bcaa4a1907c986d5744ee68c88552c407733d4d2be2"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-01-warned-couriers", "prompt": "Our risk engine sends a courier a warning after a run of orders reported as never delivered, and deactivates the account if it sees a second run. This year it has warned 987 couriers; 303 of them went on to a second run and were deactivated, and 726 are still delivering.\nSupport's view is that a warning is not a verdict. A courier whose route has a bad building collects never-arrived reports without stealing anything, and for most of the warned couriers the engine's second run never came.\nTrust & safety wants the warned-but-active couriers reviewed once, properly, instead of waiting on the engine.\n\nWork out which of them are actually stealing orders — picking the food up, marking it delivered, leaving us to refund the customer in full — and deactivate them for too many missing orders, by vendor ID. The couriers the engine already deactivated need nothing more; they are what a confirmed thief looks like.\nLeaving a thief on the road keeps costing us full refunds; deactivating an honest courier destroys what we earn from their deliveries; every decision costs a fixed review amount.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "2aa0e2a64231c77bd2e1c213aa9e9ef40d5d3fe9adb5dd18de767c69082306c1"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-01-warned-couriers-unhinted", "prompt": "Our risk engine sends a courier a warning after a run of orders reported as never delivered, and deactivates the account if it sees a second run. This year it has warned 987 couriers; 303 of them went on to a second run and were deactivated, and 726 are still delivering.\nSupport's view is that a warning is not a verdict. A courier whose route has a bad building collects never-arrived reports without stealing anything, and for most of the warned couriers the engine's second run never came.\nTrust & safety wants the warned-but-active couriers reviewed once, properly, instead of waiting on the engine.\n\nWork out which of them are actually stealing orders — picking the food up, marking it delivered, leaving us to refund the customer in full — and deactivate them for too many missing orders, by vendor ID. The couriers the engine already deactivated need nothing more.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "297d1fd2d5b6507e5e5b1fd3f44fdfc6570153c998f23e58059993a2578a6161"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-02-warned-couriers-h2", "prompt": "Our risk engine sends a courier a warning after a run of orders reported as never delivered, and deactivates the account if it sees a second run. The couriers it warned in the first half of the year have had six months for the engine to confirm or clear them; the 307 it first warned after June and never deactivated have not.\nSupport's view is that a warning is not a verdict. A courier whose route has a bad building collects never-arrived reports without stealing anything, and for most of the warned couriers the engine's second run never came.\n\nReview the couriers first warned in the second half of the year who are still delivering. Work out which of them are actually stealing orders — picking the food up, marking it delivered, leaving us to refund the customer in full — and deactivate them for too many missing orders, by vendor ID. What the engine did with the first half's warned couriers shows how a thief and an unlucky courier each look in the months after a warning.\nLeaving a thief on the road keeps costing us full refunds; deactivating an honest courier destroys what we earn from their deliveries; every decision costs a fixed review amount.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "bda89833b6bb0de2e678ac26c47d47f22ee09f6365a85f614b371ef90f779ea8"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-02-warned-couriers-h2-hard", "prompt": "Our risk engine sends a courier a warning after a run of orders reported as never delivered, and deactivates the account if it sees a second run. The couriers it warned in the first half of the year have had six months for the engine to confirm or clear them; the 307 it first warned after June and never deactivated have not.\nSupport's view is that a warning is not a verdict. A courier whose route has a bad building collects never-arrived reports without stealing anything, and for most of the warned couriers the engine's second run never came.\n\nReview the couriers first warned in the second half of the year who are still delivering. Work out which of them are actually stealing orders and deactivate them for too many missing orders, by vendor ID. What the engine did with the first half's warned couriers shows how a thief and an unlucky courier each look in the months after a warning.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "61995c2809218ed5b00ff4564c255d5fa685fabc136e5f1d94b18dd4c2c629a6"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-02-warned-couriers-h2-hardest", "prompt": "Our risk engine sends a courier a warning after a run of orders reported as never delivered, and deactivates the account if it sees a second run. The couriers it warned in the first half of the year have had six months for the engine to confirm or clear them; the 307 it first warned after June and never deactivated have not.\nSupport's view is that a warning is not a verdict. A courier whose route has a bad building collects never-arrived reports without stealing anything, and for most of the warned couriers the engine's second run never came.\n\nReview the couriers first warned in the second half of the year who are still delivering. Work out which of them are actually stealing orders and deactivate them for too many missing orders, by vendor ID.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "afd1a9d9fad2431baa16e2dee3b3cae4e35414520f2baf7127a4140a9a5f099d"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-03-farm-rings-h2", "prompt": "Trust & safety spent the first half of the year on the new-customer offer: 8,502 accounts across 396 rings were closed for farming it, every case confirmed and in the enforcement log by the end of June. Then the team was moved onto something else, and nobody has looked at the offer since.\n\nFind the accounts that farmed the new-customer offer in the second half of the year and close them for promo farming, by customer account ID. Your population is every account whose first order was placed in the second half of the year and that redeemed the offer; the accounts already in the enforcement log need no second action.\nThe second half's rings are not copies of the first half's — different people, different storefronts, some different tradecraft — but they were built the same way, and the closed cases show what a farm looks like from the inside: what its accounts shared, where they registered, what they ordered and from whom.\nClosing a farm account recovers what it would keep taking; closing a genuine new customer destroys the margin they would have brought; every decision costs a fixed review amount.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "b62f2e90b69fc5b37dd6cdab7fd615a7fef48e730d9ddb8549c5958c0487c48f"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-03-farm-rings-h2-hard", "prompt": "Trust & safety spent the first half of the year on the new-customer offer: 8,502 accounts across 396 rings were closed for farming it, every case confirmed and in the enforcement log by the end of June. Then the team was moved onto something else, and nobody has looked at the offer since.\n\nFind the accounts that farmed the new-customer offer in the second half of the year and close them for promo farming, by customer account ID. Your population is every account whose first order was placed in the second half of the year and that redeemed the offer; the accounts already in the enforcement log need no second action.\nThe second half's rings are not copies of the first half's, but they are very similar, and the closed cases show what a farm looks like from the inside.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "764698f0c2bc62f72f84cd7a704391a138911e6a11e65c3a2feaf21146896995"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-03-farm-rings-h2-hardest", "prompt": "Trust & safety spent the first half of the year on the new-customer offer: 8,502 accounts across 396 rings were closed for farming it, every case confirmed and in the enforcement log by the end of June. Then the team was moved onto something else, and nobody has looked at the offer since.\n\nFind the accounts that farmed the new-customer offer in the second half of the year and close them for promo farming, by customer account ID. Your population is every account whose first order was placed in the second half of the year and that redeemed the offer; the accounts already in the enforcement log need no second action.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "c4cad59aa0b842f65f19fc1b8f1d52788922b19e7b87dd2feb8435092aa60222"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-08", "domain": "Fraud", "id": "dsk-05a-bust-out-accounts-aug", "prompt": "Trust & safety has deactivated 2 storefronts for transaction laundering this month: 'Threshold Coffee' (vendor ID 2309445, deactivated 14 August); 'Vernon's' (vendor ID 2311144, deactivated 19 August).\nEach one ran the same play for a few weeks. The orders did not come from new accounts: they came from long-standing customers of ours whose accounts someone else had got into — each account used once or twice at that one storefront, on a card the account had never used before. Most of the owners have not noticed.\nThe storefronts are gone, but the accounts are not, and nobody believes those were the only storefronts.\n\nFind the customer accounts under the laundering rings' control — the ones behind the storefronts above, and behind any laundering storefront we have not caught yet — and lock them for account takeover, by customer account ID.\nLocking a controlled account stops what the ring would still take through it; locking a genuine customer costs us part of what they would have brought while support sorts it out; every decision costs a fixed review amount. Stolen cards are spent at hundreds of honest storefronts every month, and plenty of people change their phone or password without anything being wrong.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "16888a8d3dc7a46cfd58839fc8a1834ea0de4e6e19ad51a7691d75f408c449f5"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-08", "domain": "Fraud", "id": "dsk-05a-bust-out-accounts-aug-hard", "prompt": "Trust & safety has deactivated 2 storefronts for transaction laundering this month: 'Threshold Coffee' (vendor ID 2309445, deactivated 14 August); 'Vernon's' (vendor ID 2311144, deactivated 19 August).\nEach one ran the same play for a few weeks. The orders did not come from new accounts: they came from long-standing customers of ours whose accounts someone else had got into — each account used once or twice at that one storefront, on a card the account had never used before.\n\nFind the customer accounts under the laundering rings' control — the ones behind the storefronts above, and behind any laundering storefront we have not caught yet — and lock them for account takeover, by customer account ID.\nLocking a controlled account stops what the ring would still take through it; locking a genuine customer costs us part of what they would have brought while support sorts it out; every decision costs a fixed review amount.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "b6584a8224d54537d672944e7151d5d1e34463fbb2ead878280ad395ecfe6905"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-06-appeals-audit", "prompt": "The appeals desk overturned 2,000 courier deactivations this year, and 1,213 of those couriers are back on the road. Most appeals are against the completion-rate floor, and most of those deserve to win: a courier who declines a lot of offers is not doing anything wrong.\nCompliance sampled a handful of the reinstated accounts and found couriers who had been stealing orders before the deactivation and went back to it after: picking the food up, marking it delivered, leaving us to refund the customer in full. The desk had ruled on the reason the engine gave, not on what the courier was doing.\n\nGo through the reinstated couriers who are still active and deactivate the ones who are stealing orders, for too many missing orders, by vendor ID.\nLeaving a thief on the road keeps costing us full refunds; deactivating an honest courier destroys what we earn from their deliveries; every decision costs a fixed review amount.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "e2076d16e013f59f5c1b5e030d0dfe312f639a2b1f8b8421daf3a021181323a9"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "dsk-07-schemes-still-trading", "prompt": "Every storefront trust & safety took off the platform this year is in the enforcement log with the reason it was deactivated — 194 of them. The log also carries 1,537 other actions on storefronts: payout holds placed and released, warnings, reinstatements after an appeal. Those came to nothing.\nThe desk deactivated what it found. It did not go looking for the storefronts running the same schemes that it had not found.\n\nFind the storefronts still trading at the end of the year that are running any of the schemes the desk deactivated storefronts for, and hold their payouts pending review, by vendor ID, with the reason that fits each one.\nHolding a storefront that is running a scheme stops what it would still take; holding an honest storefront costs us a share of what we earn from it while the review runs; every review costs a fixed amount. 15,889 storefronts are still trading.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "fcdcfa264cf3aaa73d77aa82bbbb1e97ef1fd250bb1dc062298ddb7e32c86e36"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "The month the order was placed, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "Orders placed in the month that were turned away because the kitchen was too full to take them.", "name": "orders_turned_away", "type": "number"}, {"description": "Storefronts that turned away at least one order for that reason in the month.", "name": "storefronts", "type": "number"}], "description": "For every month of 2024, the orders the app turned away because the kitchen was too full to take them, and the storefronts that turned any away.", "key": ["month"], "max_rows": 5000, "name": "kitchen_full_turnaways_monthly_2024", "order_by": ["month"], "row_count": 12, "title": "Merchant operations › Kitchen-full turn-aways, 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "sig-01-kitchen-full-turnaways-hard", "prompt": "Merchant operations needs us to rebuild their dashboard's data sources based on the new reporting schema.\nThe kitchen capacity tile shows how often restaurants were too busy to take an order during 2024. When a kitchen is too full to get a new order out in time, the order is turned away and the customer is told the restaurant is too busy to take it. Rebuild the tile from the warehouse and publish it.\n\nAn order counts in the month it was placed.\n\nPublish the merchant operations data source representing orders turned away by full kitchens in each month of 2024, at a monthly grain, ascending by month (`kitchen_full_turnaways_monthly_2024`):\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the order was placed, as YYYY-MM. |\n| `orders_turned_away` | number | Orders placed in the month that were turned away because the kitchen was too full to take them. |\n| `storefronts` | number | Storefronts that turned away at least one order for that reason in the month. |", "version": "2ab7d6f059f1944e958e4eb4873e2d120d0b40e50d47b5f50e807126a26201fd"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-12", "domain": "Operations Research", "id": "smp-01-courier-location-report-load", "prompt": "Platform engineering is replacing the service that takes in courier location reports and wants\nto size the new one on last year's real traffic. The courier app reports its location whenever\nsomething happens (signing on, accepting a job, reaching the restaurant, finishing a drop-off,\nbeing taken off a job), and on top of that it checks in every half hour while a courier is\nonline with no job.\nHow many location reports did the app send in each month of 2024, New York time? File them as\n12 metrics, reports_2024_01 through reports_2024_12.", "version": "c00433323f3e0a63f98ab6863d17ddc453078bea4a16fbd9db00f5a822e69c6a"} {"area": "Growth, Marketing & Membership", "contracts": {"data_sources": [{"columns": [{"description": "Calendar-day follow-up length.", "name": "age_days", "type": "integer", "values": [30, 60, 90]}, {"description": "Customers initially assigned an acquisition offer.", "name": "offer_customers", "type": "integer"}, {"description": "Customers initially assigned the offer holdout.", "name": "holdout_customers", "type": "integer"}, {"description": "Mean customer contribution in the offer arm.", "name": "offer_contribution_per_customer_usd", "type": "number"}, {"description": "Mean customer contribution in the holdout arm.", "name": "holdout_contribution_per_customer_usd", "type": "number"}, {"description": "Offer mean minus holdout mean.", "name": "incremental_contribution_per_customer_usd", "type": "number"}, {"description": "Lower endpoint of the specified 80% confidence interval.", "name": "ci80_lower_usd", "type": "number"}, {"description": "Upper endpoint of the specified 80% confidence interval.", "name": "ci80_upper_usd", "type": "number"}], "key": ["age_days"], "max_rows": 5000, "name": "membership_acquisition_payback", "order_by": ["age_days"], "row_count": 3, "title": "Membership offers: customer contribution payback"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "pe-01-membership-acquisition-payback", "prompt": "Did Q1 membership acquisition offers pay for themselves within 90 days? Publish incremental platform contribution per customer initially targeted, at 30, 60, and 90 days, relative to the contemporaneous offer holdout, with 80% confidence intervals.\n\nUse each customer's first recorded checkout offer opportunity in Q1 2024, restricted to customers who had no membership contract before that opportunity. Follow the initial assignment under the continuing targeting policy. Each window includes the opportunity's New York calendar date and excludes that date plus the stated number of days. Compare the simple customer means of the two initial arms; use an unpooled normal interval for their difference, with sample variances across customers and z=1.2815515655446004.\n\nUse Finance's final booked contribution on orders placed in that window. Use the numbers as Finance booked them against each order. Count every order placed in the month (by ORDERED_DATE), whatever happened to it afterwards: cancelled orders count too.\nFor revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order.\nFor costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery.\nAt customer level, subscription earnings must be counted once, including earnings retained outside orders, and pickup membership credits on those orders are an expense when issued. For this calendar-day cohort view, collected monthly fees are earned on capture; collected annual fees are earned in twelve installments at 30-day intervals after capture, the first at day 30. Round each installment to cents and put the residual in the twelfth. Use actual fees net of refunds. Report per-customer results to six decimal places.\n\nPublish the data source `membership_acquisition_payback, one row per (`age_days`), `age_days` ascending:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `age_days` | integer | Calendar-day follow-up length. One of: `30`, `60`, `90`. |\n| `offer_customers` | integer | Customers initially assigned an acquisition offer. |\n| `holdout_customers` | integer | Customers initially assigned the offer holdout. |\n| `offer_contribution_per_customer_usd` | number | Mean customer contribution in the offer arm. |\n| `holdout_contribution_per_customer_usd` | number | Mean customer contribution in the holdout arm. |\n| `incremental_contribution_per_customer_usd` | number | Offer mean minus holdout mean. |\n| `ci80_lower_usd` | number | Lower endpoint of the specified 80% confidence interval. |\n| `ci80_upper_usd` | number | Upper endpoint of the specified 80% confidence interval. |", "version": "d8c09f2e11ff44703ea9f29d943193beceee544d2b025753403fe4559b424625"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "New York day-end.", "name": "day", "type": "date"}, {"description": "Lower bound on positive cohort funds still withheld at day-end.", "name": "withheld_lower_usd", "type": "number"}, {"description": "Upper bound on positive cohort funds still withheld at day-end.", "name": "withheld_upper_usd", "type": "number"}, {"description": "Lower bound on positive cohort funds authorized but not received at day-end.", "name": "in_transit_lower_usd", "type": "number"}, {"description": "Upper bound on positive cohort funds authorized but not received at day-end.", "name": "in_transit_upper_usd", "type": "number"}, {"description": "Cumulative cohort funds received by day-end.", "name": "received_usd", "type": "number"}, {"description": "Lower bound on cumulative sum of day-end withheld balances.", "name": "withheld_lower_dollar_days", "type": "number"}, {"description": "Upper bound on cumulative sum of day-end withheld balances.", "name": "withheld_upper_dollar_days", "type": "number"}, {"description": "Lower bound on cumulative sum of day-end in-transit balances.", "name": "in_transit_lower_dollar_days", "type": "number"}, {"description": "Upper bound on cumulative sum of day-end in-transit balances.", "name": "in_transit_upper_dollar_days", "type": "number"}], "key": ["day"], "max_rows": 5000, "name": "merchant_held_payouts_daily", "order_by": ["day"], "row_count": 184, "title": "Merchant payout holds and transfers"}, {"columns": [{"description": "Rank by lower-bound withheld dollar-days descending, invoice ID ascending.", "name": "rank", "type": "integer"}, {"description": "Payout run invoice.", "name": "invoice_id", "type": "integer"}, {"description": "Merchant supplier identifier.", "name": "vendor_id", "type": "integer"}, {"description": "The run's own positive net payout.", "name": "payout_usd", "type": "number"}, {"description": "Lower bound on sum of day-end withheld balances through year-end.", "name": "withheld_lower_dollar_days", "type": "number"}, {"description": "Upper bound on sum of day-end withheld balances through year-end.", "name": "withheld_upper_dollar_days", "type": "number"}, {"description": "Lower bound on sum of day-end in-transit balances through year-end.", "name": "in_transit_lower_dollar_days", "type": "number"}, {"description": "Upper bound on sum of day-end in-transit balances through year-end.", "name": "in_transit_upper_dollar_days", "type": "number"}, {"description": "Withheld plus in-transit balance at year-end.", "name": "unreceived_at_year_end_usd", "type": "number"}], "key": ["invoice_id"], "max_rows": 5000, "name": "merchant_held_payouts_top20", "order_by": ["rank"], "row_count": 20, "title": "Merchant runs with the largest accumulated hold delay"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "pe-04-merchant-payout-dollar-days", "prompt": "For positive merchant payout runs first placed on hold in Q3 2024, publish a daily view through year-end of amounts withheld and amounts released but not yet received by the merchant, together with the cumulative amount received. Give lower and upper bounds for the waiting balances and their cumulative dollar-days, and show the twenty payout runs with the largest lower bound on accumulated withheld dollar-days.\n\nUse every New York day-end from July 1 through December 31. For this daily dashboard a dollar-day is one dollar in a waiting state at a day-end, summed across days. Keep each run's own net payable and partial payments; later recovery adjustments belong to the runs on which they were booked. Bound authorization dates using only hold placement, recorded receipt dates and the merchant transfer policy: receipt is one business day after authorization, with business days Monday through Friday. An authorization may occur on any calendar day, never before placement. Assume every authorization is represented by a payment in this complete extract. Rank runs by lower-bound withheld dollar-days descending, breaking ties by invoice identifier ascending. USD amounts are to the cent.\n\nPublish the two following data sources: \n`merchant_held_payouts_daily`: Merchant payout holds and transfers\n\n| column | type | meaning |\n| --- | --- | --- |\n| `day` | date | New York day-end. |\n| `withheld_lower_usd` | number | Lower bound on positive cohort funds still withheld at day-end. |\n| `withheld_upper_usd` | number | Upper bound on positive cohort funds still withheld at day-end. |\n| `in_transit_lower_usd` | number | Lower bound on positive cohort funds authorized but not received at day-end. |\n| `in_transit_upper_usd` | number | Upper bound on positive cohort funds authorized but not received at day-end. |\n| `received_usd` | number | Cumulative cohort funds received by day-end. |\n| `withheld_lower_dollar_days` | number | Lower bound on cumulative sum of day-end withheld balances. |\n| `withheld_upper_dollar_days` | number | Upper bound on cumulative sum of day-end withheld balances. |\n| `in_transit_lower_dollar_days` | number | Lower bound on cumulative sum of day-end in-transit balances. |\n| `in_transit_upper_dollar_days` | number | Upper bound on cumulative sum of day-end in-transit balances. |\n\n- **Grain:** one row per (`day`).\n- **Order:** `day` ascending.\n- **Rows:** exactly 184.\n\nData source `merchant_held_payouts_top20`: Merchant runs with the largest accumulated hold delay\n\n| column | type | meaning |\n| --- | --- | --- |\n| `rank` | integer | Rank by lower-bound withheld dollar-days descending, invoice ID ascending. |\n| `invoice_id` | integer | Payout run invoice. |\n| `vendor_id` | integer | Merchant supplier identifier. |\n| `payout_usd` | number | The run's own positive net payout. |\n| `withheld_lower_dollar_days` | number | Lower bound on sum of day-end withheld balances through year-end. |\n| `withheld_upper_dollar_days` | number | Upper bound on sum of day-end withheld balances through year-end. |\n| `in_transit_lower_dollar_days` | number | Lower bound on sum of day-end in-transit balances through year-end. |\n| `in_transit_upper_dollar_days` | number | Upper bound on sum of day-end in-transit balances through year-end. |\n| `unreceived_at_year_end_usd` | number | Withheld plus in-transit balance at year-end. |\n\n- **Grain:** one row per (`invoice_id`).\n- **Order:** `rank` ascending.\n- **Rows:** exactly 20.", "version": "eabc10321e10d7a21cb5bc4f0bd2a27deb2390ae705dae91c91538cd86736b41"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-08", "domain": "Fraud", "id": "now-02-accounts-worked-by-someone-else", "prompt": "It's the end of August 2024. Some couriers let somebody else work their account: they rent it out, split the day with someone, or hand it to a broker who runs a crew across several accounts. The person on the bike never passed our checks. These arrangements come and go, and plenty of the accounts that were being worked by someone else earlier this year are back with their holder. Work out which accounts are being worked by somebody else right now and deactivate them for identity farming, carrying their vendor ID. A courier on a new phone, or sharing a phone with someone at home who also delivers, is not renting.", "version": "eff593f6217f5d91624fcc1ae339e7a18ed052489f5514eadb7460b9b379e22f"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-08", "domain": "Fraud", "id": "cnc-02-cancellation-farming-august", "prompt": "It is the end of August 2024. Customers, couriers and storefronts all cancel orders for ordinary reasons: an item runs out, a special request can't be met, a courier can't find the door. Some of the cancelling is a scheme. Identify the customers, couriers or businesses whose cancellation activity is suspicious and ban them, using their vendor ID or customer account ID.", "version": "98cb54d2dcd026b44c4b0924ac349791bffd73c46b2108f9b9e0b67183e8f3d1"} {"area": "Growth, Marketing & Membership", "contracts": {"data_sources": [{"columns": [{"description": "The New York calendar day, YYYY-MM-DD.", "name": "day", "type": "date"}, {"description": "Hours couriers spent on shift during the day, the whole fleet together.", "name": "courier_hours", "type": "number"}, {"description": "Average number of couriers on shift at any given moment of the day.", "name": "avg_couriers_on_shift", "type": "number"}], "description": "For every day of the fourth quarter of 2024, the hours couriers spent on shift during that New York day.", "key": ["day"], "max_rows": 5000, "name": "courier_shift_hours_daily_q4_2024", "order_by": ["day"], "row_count": 92, "title": "Courier operations › Time on shift by day, Q4 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "clk-02-courier-shift-hours-q4", "prompt": "Courier operations needs the data source behind its daily supply line rebuilt on the new reporting schema: the hours couriers spent on shift on each day of the fourth quarter of 2024, and how many couriers that meant were on shift at a time.\n\nA courier is on shift from logging on until logging off, whether or not a job comes in. Days are New York calendar days, and each day gets only the time that fell inside it.\n\nPublish `courier_shift_hours_daily_q4_2024`, one row per day from 1 October to 31 December 2024, ascending by day:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `day` | date | The New York calendar day, YYYY-MM-DD. |\n| `courier_hours` | number | Hours couriers spent on shift during the day, the whole fleet together. |\n| `avg_couriers_on_shift` | number | Average number of couriers on shift at any given moment of the day. |", "version": "b9a3a5a72c466830f287ddc9eae5f65c7f890bcdfbe8a59df8a4fb9cbed16895"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "The borough of the order's pickup zone, as the marketplace tables number boroughs (1 to 5).", "name": "borough_id", "type": "integer", "values": [1, 2, 3, 4, 5]}, {"description": "Delivery orders placed in the month and picked up in the borough, every status.", "name": "delivery_orders", "type": "integer"}, {"description": "The minimum-pay true-up Finance accrued against those orders divided by `delivery_orders`, in dollars per delivery.", "name": "min_pay_true_up_per_delivery_usd", "type": "number"}], "description": "Monthly minimum-pay true-up accrued per delivery order placed, by the borough the order was picked up in, for every month of 2024.", "key": ["month", "borough_id"], "max_rows": 5000, "name": "minimum_pay_true_up_monthly_per_delivery_by_borough_2024", "order_by": ["month", "borough_id"], "row_count": 60, "title": "Finance › Minimum-pay true-up per delivery by borough, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-36t-min-pay-true-up-per-delivery-by-borough", "prompt": "Finance needs the data source behind its minimum-pay tile rebuilt on the new reporting schema: what the minimum-pay rule cost us per delivery on each month's orders, by the borough the order was picked up in, for 2024.\n\nIn April the platform started to comply with the DCWP minimum-pay rule, and whenever a courier's pay for a week falls short of the minimum the platform tops it up with a true-up. The tile carries the true-up as Finance accrued it against each delivery, not the week's top-up spread evenly. Every delivery order placed in the month counts, whatever happened to it afterwards; pickup orders are not deliveries.\n\nPublish `minimum_pay_true_up_monthly_per_delivery_by_borough_2024` at a monthly grain, ascending by month and then borough:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `borough_id` | integer | The borough of the order's pickup zone, as the marketplace tables number boroughs (1 to 5). |\n| `delivery_orders` | integer | Delivery orders placed in the month and picked up in the borough, every status. |\n| `min_pay_true_up_per_delivery_usd` | number | The minimum-pay true-up Finance accrued against those orders divided by `delivery_orders`, in dollars per delivery. |", "version": "2c026b985d3b445fa72e330ea76fd7350b533bd66833ebc28ca3c3ba2f6a8aba"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "The month the orders were placed in, as YYYY-MM.", "name": "month", "type": "month"}, {"description": "The borough of the order's pickup zone, as the marketplace tables number boroughs (1 to 5).", "name": "borough_id", "type": "integer", "values": [1, 2, 3, 4, 5]}, {"description": "Orders placed in the month and picked up in the borough, every status.", "name": "orders", "type": "integer"}, {"description": "What Finance booked as paid to storefronts for the period's orders cancelled because no courier came, divided by `orders`, in dollars per order.", "name": "no_courier_compensation_per_order_usd", "type": "number"}], "description": "Monthly storefront compensation on orders cancelled for want of a courier, per order placed, by the borough the order was picked up in, for every month of 2024.", "key": ["month", "borough_id"], "max_rows": 5000, "name": "no_courier_compensation_monthly_per_order_by_borough_2024", "order_by": ["month", "borough_id"], "row_count": 60, "title": "Finance › Storefront compensation for no-courier cancellations per order by borough, monthly 2024"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "ds-40t-no-courier-compensation-by-borough", "prompt": "Finance needs the data source behind its cancellations tile rebuilt on the new reporting schema: what we paid storefronts for orders cancelled because no courier came for them, per order, on each month's orders, by the borough the order was picked up in, for 2024.\n\nThe compensation is what Finance booked as paid to the storefront on an order cancelled for want of a courier; cancellations for any other reason are not part of it. Every order placed in the month counts, whatever happened to it afterwards.\n\nPublish `no_courier_compensation_monthly_per_order_by_borough_2024` at a monthly grain, ascending by month and then borough:\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | The month the orders were placed in, as YYYY-MM. |\n| `borough_id` | integer | The borough of the order's pickup zone, as the marketplace tables number boroughs (1 to 5). |\n| `orders` | integer | Orders placed in the month and picked up in the borough, every status. |\n| `no_courier_compensation_per_order_usd` | number | What Finance booked as paid to storefronts for the period's orders cancelled because no courier came, divided by `orders`, in dollars per order. |", "version": "30b312bc99770549691f5ff332dbc9afaba8664f723f0ed4cf48717964e540fc"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "cb-01p-december-lost-disputes-vs-reserve", "prompt": "How much do we expect to lose on card disputes that resolve in December 2024? Finance needs a forecast for the month.\n\nFile one `lost_disputes_usd` forecast, with an 80% central prediction interval, for the amounts for lost disputes, assigning them to months by their resolution date. Include disputes over earlier orders or opened in earlier months if they resolve in December.\n\nFinance's reserve for December is $225,500: the trailing three months' average of lost disputes, grossed up for December's order volume.\n\nFinance will hold your forecast against that figure, so file your own view of December, not the reserve.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "1856214b96ed47570f0faf7ad68c350b6f3ebe0f1b0091670849fa3654a599bf"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "mr-01p-december-membership-revenue-vs-plan", "prompt": "How much membership revenue do we expect to recognize in December 2024? Finance needs the revenue on the books for the month, not the cash coming in from renewals.\n\nFile one `membership_revenue_usd` forecast, with an 80% central prediction interval, for the membership revenue posted to the books in December.\n\nFP&A's plan books $2.418 million for December: November's posting, grossed up for the year-end renewal cohort.\n\nFP&A will hold your forecast against that figure, so file your own view of December, not the plan.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "57c41150b6248752d8fd30d28959badefd24a13b8abcffd17859db63d519d104"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "sw-12p-december-negotiable-checks-vs-estimate", "prompt": "We're preparing for the December close. How many December-dated AP checks will still be negotiable in the December 31 extract, and how much will they total?\n\nFile two forecasts, each with an 80% central prediction interval: `december_negotiable_checks_count` and `december_negotiable_checks_usd`. A check belongs to its check date, and counts if it has not cleared by the close.\n\nTreasury's estimate for the close is about 26,500 checks and $10.4 million: the last week of November's issue run-rate against the clearing lag it sees on the bank statement.\n\nTreasury will hold your forecast against that figure, so file your own view of the close, not the estimate.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "c29e107fd604969a38dbb29cf8a3d1f36c53ea0cb55508379e9a24399a95a7dc"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "col-22-phantom-search-1", "prompt": "Finance is reviewing missing-item refunds at the storefronts that took at least\n1,500 and fewer than 15,000 orders in 2024, not counting cancelled orders.\n\nUnder the merchant error-charge policy a storefront is billed for the items it leaves out\nof the bag, scaled by its liability tier, so at storefronts on the most favourable tiers the\nplatform ends up funding most or all of every missing-item refund. Some storefronts have been\ncaught working with a few of their own regulars: every few orders an item goes \"missing\", we\nrefund it, the storefront barely sees a deduction, and the customer keeps ordering there.\n\nFind the storefronts in that range that were doing this in 2024, put their payouts on hold for\nrefund collusion, and close the customer accounts that were in on it.\n\nA ring left trading and an honest storefront put on hold cost us the same. The storefronts\nare two thirds of the review; the customer closures are the other third, judged on how many of\nthe rings' accounts you close and how few honest customers you close with them. Busy storefronts collect repeat claims from ordinary\ncustomers, and a kitchen can short the same loyal customer more than once.", "prompt_revision": "corrected after the final run: the prompt now states how the grader weighs the review (the storefronts two thirds, the customer closures one third), and the question was run again on this wording", "version": "ee7ee255bd1b63cd76443bb2c771f3077c1da9a6beaa7062c23115893e67ef1b"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "col-22-phantom-search-2", "prompt": "Finance is reviewing missing-item refunds at the storefronts that took at least\n1,500 and fewer than 45,000 orders in 2024, not counting cancelled orders.\n\nUnder the merchant error-charge policy a storefront is billed for the items it leaves out\nof the bag, scaled by its liability tier, so at storefronts on the most favourable tiers the\nplatform ends up funding most or all of every missing-item refund. Some storefronts have been\ncaught working with a few of their own regulars: every few orders an item goes \"missing\", we\nrefund it, the storefront barely sees a deduction, and the customer keeps ordering there.\n\nFind the storefronts in that range that were doing this in 2024, put their payouts on hold for\nrefund collusion, and close the customer accounts that were in on it.\n\nA ring left trading and an honest storefront put on hold cost us the same. The storefronts\nare two thirds of the review; the customer closures are the other third, judged on how many of\nthe rings' accounts you close and how few honest customers you close with them. Busy storefronts collect repeat claims from ordinary\ncustomers, and a kitchen can short the same loyal customer more than once.", "prompt_revision": "corrected after the final run: the prompt now states how the grader weighs the review (the storefronts two thirds, the customer closures one third), and the question was run again on this wording", "version": "9dc415eb85b1e41e4a470b8c1c4e2966ebe42418acebebd3239815c3247840de"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "col-22-phantom-search-3", "prompt": "Finance is reviewing missing-item refunds at every storefront on the platform in 2024.\n\nUnder the merchant error-charge policy a storefront is billed for the items it leaves out\nof the bag, scaled by its liability tier, so at storefronts on the most favourable tiers the\nplatform ends up funding most or all of every missing-item refund. Some storefronts have been\ncaught working with a few of their own regulars: every few orders an item goes \"missing\", we\nrefund it, the storefront barely sees a deduction, and the customer keeps ordering there.\n\nFind the storefronts on the platform that were doing this in 2024, put their payouts on hold for\nrefund collusion, and close the customer accounts that were in on it.\n\nA ring left trading and an honest storefront put on hold cost us the same. The storefronts\nare two thirds of the review; the customer closures are the other third, judged on how many of\nthe rings' accounts you close and how few honest customers you close with them. Busy storefronts collect repeat claims from ordinary\ncustomers, and a kitchen can short the same loyal customer more than once.", "prompt_revision": "corrected after the final run: the prompt now states how the grader weighs the review (the storefronts two thirds, the customer closures one third), and the question was run again on this wording", "version": "db7ca0355ad431622058c1f00a8216443e6e1c74e58d1612dfc7603fa8518f21"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "col-22-phantom-search-4", "prompt": "Finance is reviewing missing-item refunds at every storefront on the platform in 2024.\n\nSome storefronts have been caught working with a few of their own customers to collect\nmissing-item refunds on bags that were never short.\n\nFind the storefronts on the platform that were doing this in 2024, put their payouts on hold for\nrefund collusion, and close the customer accounts that were in on it.\n\nA ring left trading and an honest storefront put on hold cost us the same. The storefronts\nare two thirds of the review; the customer closures are the other third, judged on how many of\nthe rings' accounts you close and how few honest customers you close with them.", "prompt_revision": "corrected after the final run: the prompt now states how the grader weighs the review (the storefronts two thirds, the customer closures one third), and the question was run again on this wording", "version": "40b4bedb97653193d7733b815b4486c113f8b6b135b466325917d8ff1fe08ae1"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "gps-01-location-spoofing", "prompt": "Dispatch thinks some couriers are telling us they are somewhere they are not,\nso the offers go to them instead of whoever is actually closest. Work out which ones and\ntake them off the platform.\n\nBan couriers with a call carrying the `VENDOR_ID` of every courier you are removing, for\nreason `Reason.GPS_SPOOFING`.", "version": "ccc0896cb364704a41389a016561fe366adfd22d8a27c0ac062c85c62475605d"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "gps-01-location-spoofing-giveaway", "prompt": "Dispatch thinks some couriers are telling us they are somewhere they are not,\nso the offers go to them instead of whoever is actually closest. Work out which ones and\ntake them off the platform.\n\nWhat the closed cases looked like: the handset reports a position the courier could not\nhave reached from where it last reported, and the legs it was dispatched for do not line up\nwith where the phone says it went. The platform's own mock-location flag was on for most of\nthem, and for a great many honest couriers too.\n\nBan couriers with a call carrying the `VENDOR_ID` of every courier you are removing, for\nreason `Reason.GPS_SPOOFING`.", "version": "8179683d9ce08b05be64d31e076ca402435e1ba141777071b8cb1028e1392b21"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "bot-01-order-grab-automation", "prompt": "Couriers are complaining that the good offers disappear before they can\naccept them. Work out which couriers are taking offers with software rather than by hand,\nand ban them from the platform.\n\nBan couriers with a call carrying the `VENDOR_ID` of every courier you are removing, for\nreason `Reason.OTHER`.", "version": "311b69690bf92ba977b4e3bb2034e542af9296e1dd11a4103c41c429d4e52bcc"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "bot-01-order-grab-automation-giveaway", "prompt": "Couriers are complaining that the good offers disappear before they can\naccept them. Work out which couriers are taking offers with software rather than by hand,\nand ban them from the platform.\n\nWhat the closed cases looked like: the time between an offer going out and the answer\ncoming back, over and over, is shorter than a person can read and tap. We already caught a\nfew of them this way — see the automation entries in the violations log — but the ones we\ncaught were the careless ones.\n\nBan couriers with a call carrying the `VENDOR_ID` of every courier you are removing, for\nreason `Reason.OTHER`.", "version": "09066893f72b546c5da6be6eeda50091ad90532d50d850595093e982d687d05a"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "ids-01-identity-substitution", "prompt": "Trust and safety thinks some accounts have been compromised and are being \noperated by someone other than the courier who originally signed up. Work out which accounts \nand take them off the platform.\n\nBan couriers with a call carrying the `VENDOR_ID` of every courier you are removing, for\nreason `Reason.IDENTITY_FARMING`.", "version": "d3c73e852f8c8881abe0dae76f3641d3ef239ce31ad998cf273d7faa8eb5b9bc"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "ids-01-identity-substitution-giveaway", "prompt": "Trust and safety thinks some accounts have been compromised and are being \noperated by someone other than the courier who originally signed up. Work out which accounts \nand take them off the platform.\n\nWhat the closed cases looked like: the account is asked to prove who is holding the phone\nmore often than most, and when a reviewer looks at the selfie it is not the same person.\nThe account is also usually seen on more than one handset. Note that a courier we already\ndeactivated, working somebody else's live account, fails the same check for a different\nreason — that is a separate review.\n\nBan couriers with a call carrying the `VENDOR_ID` of every courier you are removing, for\nreason `Reason.IDENTITY_FARMING`.", "version": "a9127f5fa91ec9e1d096d8797cd0268d651d43ff7e299b6512ad98ad995bbbad"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "vbp-01-duplicate-brands", "prompt": "Search is complaining that a single restaurant is making multiple profiles, \ntaking up multiple result spots when customers search. We think\nsome operators have listed the same kitchen multiple times under invented brand names, so\nthey hold several slots where they should hold one.\n\nWork out which storefronts are part of that and take them off the platform.\n\nBan merchants with a call carrying the `VENDOR_ID` of every storefront you are removing,\nfor reason `Reason.GHOST_KITCHEN`.", "version": "08a8eb67431dd05ea764f9f96be38e1527c6a7e7baa9fda5e660079518a976cf"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "vbp-01-duplicate-brands-giveaway", "prompt": "Search is complaining that a single restaurant is making multiple profiles, \ntaking up multiple result spots when customers search. We think\nsome operators have listed the same kitchen multiple times under invented brand names, so\nthey hold several slots where they should hold one.\n\nWork out which storefronts are part of that and take them off the platform.\n\nWhat the closed cases looked like: every brand in a group trades from the same street\naddress, sells the same menu, and the names are variations on one word. Be careful — plenty\nof honest storefronts share an address too, because food halls and shared commissary\nkitchens really do put a dozen separate businesses at one door.\n\nBan merchants with a call carrying the `VENDOR_ID` of every storefront you are removing,\nfor reason `Reason.GHOST_KITCHEN`.", "version": "8d9aa181f8d1a89525cda62e16c332ed40cc0ea16ffc78493c68f6f79368ac82"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "pay-01-diverted-payouts", "prompt": "Merchant support has been restoring payout bank accounts for storefronts that\nsay they never changed them. Finance wants to first confirm the size of these payouts.\n\nWork out how much of our 2024 merchant payout money went to a bank account the storefront\ndid not own, and how much of that we have already refunded.\n\nReport with `report_metrics,` unit `usd,` using these names exactly:\n\n* `diverted_usd,` payouts that went to an account the storefront had not authorised.\n* `made_whole_usd,` of that, what we have since restored to the storefront.\n* `outstanding_usd,` diverted_usd` minus `made_whole_usd`.\n\nAnd one figure with unit `count`:\n\n* `affected_merchants,` how many storefronts had payouts diverted.\n", "version": "95ecbdee39e3b5a9b29cff07558694a7288aaa6fd26fd4e2fd950aea36972a5c"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "fc-10t2-july-true-ups-with-holiday-plan", "prompt": "In April the platform started to comply with the DCWP minimum-pay rule for food delivery workers: $19.56 an hour on the time couriers are online, topped up weekly with a true-up where a week falls short. At the same time the dispatcher started restricting when couriers can go online, so the online time we pay for is busier. From 1 July it tightens further: about 0.49 connected hours per forecast order (0.69 today), and its plan for the payout periods ending in July is about 2.09 million connected hours in total, against about 2.75 million for the periods ending in June. Volume is planned flat, except the Independence Day week (1 to 7 July), which marketing expects to run about a tenth below a normal week.\n\nFile one `minimum_pay_true_ups_usd` forecast, with an 80% central prediction interval, for the sum of the adjustments on the weekly courier pay invoices for the payout periods ending in July. A payout period belongs to the month its end date falls in; tips are not pay.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "ccd918f55178e8f70edcb908174bd9f72060e64bc69ecfaa2e02bf5d9d1db65f"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-06", "domain": "Forecasting", "id": "fc-10t3-july-true-ups-weekly-plan", "prompt": "In April the platform started to comply with the DCWP minimum-pay rule for food delivery workers: $19.56 an hour on the time couriers are online, topped up weekly with a true-up where a week falls short. At the same time the dispatcher started restricting when couriers can go online, so the online time we pay for is busier. From 1 July it tightens further: about 0.49 connected hours per forecast order (0.69 today), and its plan for the payout periods ending in July is about 520,000 connected hours a week, the same every week, or about 2.09 million over the four periods (the five periods ending in June came to about 2.75 million). Volume is planned flat, except the Independence Day week (1 to 7 July), which marketing expects to run about a tenth below a normal week.\n\nFile one `minimum_pay_true_ups_usd` forecast, with an 80% central prediction interval, for the sum of the adjustments on the weekly courier pay invoices for the payout periods ending in July. A payout period belongs to the month its end date falls in; tips are not pay.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "e70845b7178ae3e16ee6950d739a98354b65520c06926ee9267059c0eac0204c"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "gm-01-unpaid-renewal-benefits", "prompt": "Membership can continue to waive fees while a renewal is being collected. Finance wants a\nforecast of the fee concessions attached to June monthly-plan bills that will still be\nunpaid when their contractual payment grace ends.\n\nFor those bills, measure delivery and service fees waived on non-cancelled orders before\ncollection, within the billed service period and the first 30 calendar days from the\noriginal bill date. Exclude the collection date itself. A bill paid on the last grace\ndate counts as paid in time. Count each bill once, however many collection attempts it\ntakes. The cohort is the month of the original bill, and the 30-day window includes the\nbill date and excludes that date plus 30 days. Use New York dates.\n\nFile `unpaid_renewal_benefits_usd` for June 2024 with an 80% central prediction interval.", "version": "80bac01dc854da8b0ef3372ecbb29a77f5cfe07f13b91b0ec0d36a29547f03ea"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-08", "domain": "Forecasting", "id": "gm-06-membership-trial-90d-value", "prompt": "Growth needs a forecast of the platform contribution\nthe September trial cohort will generate in its first 90 days.\n\nUse each customer's first recorded membership trial start. Assign customers to that\nstart's month, and measure contribution over the New York start date and the next 89\ncalendar dates. Include every customer starting a trial, including those who place no\norders or never become paying members. Report mean contribution per customer in the\ncohort. This is the cohort's observed economics, not an estimate of the causal effect of\noffering a trial.\n\nUse Finance's final booked order contribution: merchant commission and the delivery,\nservice, small-order and regulatory response fees we keep, less courier pay, our share of\npromotions and refunds, processing, chargebacks, cancellation costs, referral and quest\nincentives, and the minimum-pay true-up accrued to the delivery. Include every order\nstatus. The later payout and reversal of a true-up do not replace its order accrual.\n\nAt customer level, count subscription earnings once, including earnings retained outside\norders. Remove the subscription fee allocated to orders before adding the customer's\nearned fees, and expense membership pickup credits when issued. For this 90-day cohort\nview, monthly fees are earned on capture; annual fees are earned in twelve installments\nat 30-day intervals after capture, starting at day 30. Round installments to cents and\nput the residual in the twelfth. Use actual fees net of refunds.\n\nFile `trial_customer_90d_contribution_usd` for September 2024 with an 80% central prediction interval.", "version": "6dd464f61aea861d4aee0cd5c78355557c9ddb81939ba5d49b0f295f4ad749d8"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "gm-03-promotional-acquisition-value", "prompt": "Growth wants a June forecast for the first 30 days of platform contribution per new\ncustomer acquired with a platform-funded coupon. A customer enters this cohort when their\nfirst recorded order carries a coupon discount funded at least partly by us. This is a\ndescription of those customers, not a claim that the coupon caused their acquisition.\n\nAssign the customer to the month of that first order. Include all of their orders from\nthat New York calendar date through the next 29 dates, whether or not they use another\ncoupon or join membership. Divide total customer contribution by every customer in the\ncohort, including customers who never order again.\n\nUse Finance's final booked order contribution: merchant commission and the delivery,\nservice, small-order and regulatory response fees we keep, less courier pay, our share of\npromotions and refunds, processing, chargebacks, cancellation costs, referral and quest\nincentives, and the minimum-pay true-up accrued to the delivery. Include every order\nstatus. The later payout and reversal of a true-up do not replace its order accrual.\n\nAt customer level, count subscription earnings once, including earnings retained outside\norders. Remove the subscription fee allocated to orders before adding the customer's\nearned fees, and expense membership pickup credits when issued. For this 30-day cohort\nview, monthly fees are earned on capture; annual fees are earned in twelve installments\nat 30-day intervals after capture, starting at day 30. Round installments to cents and\nput the residual in the twelfth. Use actual fees net of refunds.\n\nFile `acquired_customer_30d_contribution_usd` for June 2024 with an 80% central prediction interval.", "version": "4abe11e09c77fd5ef8196e125ad691d279c7c7599db06da61674bcbbcda36b0f"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-04", "domain": "Forecasting", "id": "gro-01-new-paid-memberships-may", "prompt": "Membership growth needs a number for May: how many memberships will convert to a paid plan during the month? A conversion is a free trial ending with the customer staying on as a paying member, or a customer taking a paid plan at checkout, including customers coming back after a lapse; renewals of members already paying do not count. Two things changed this year. In February we introduced a 90-day free trial for customers who had turned down the monthly plan more than once, alongside the standard 30-day trial; and on 1 March we started offering the annual plan at half price at checkout to customers who decline the monthly plan.\n\nBased on what you know and the numbers from January to April, file a forecast for the number of memberships converting to a paid plan (`new_paid_memberships`) in May 2024, with an 80% central prediction interval.", "version": "997fbe74eb31c4e31f20453b0246b78755262747fedae1a86a8ee51538713890"} {"area": "Accounting, Close & Regulatory Reporting", "data_cutoff": "2024-02", "domain": "Forecasting", "id": "mem-01-annual-launch-fees-recognised-june", "prompt": "Our membership has two paid plans, a monthly plan at $9.99 for 30 days and an annual plan at $96 for 365 days, plus free trials and partner-sponsored memberships that pay nothing. On 1 March we started offering the annual plan at half price, $48 for the first year, to customers who turn down the monthly plan at checkout; the monthly plan, the trials and renewals are unchanged. Growth expects the half-price offer to bring in about 5 thousand new annual members in March, 12 thousand in April, 21 thousand in May and 25 thousand in June, on top of the annual renewals already falling due.\n\nFinance recognises a monthly fee in full on the day the card is charged successfully, not the day the bill is raised, so a bill that fails and is recovered on a retry counts on the day the retry succeeds. An annual fee is not revenue when it is charged: it goes to deferred membership revenue and is released to revenue in twelve equal steps, one every 30 days after the charge.\n\nBased on what you know, the numbers from January and February and that plan, file a forecast for the membership fee revenue we will recognise (`membership_fees_recognised_usd`) in June 2024, with an 80% central prediction interval, counting every fee dollar recognised once, whether Finance leaves it in membership revenue or credits it to the member's orders.", "version": "17d0344d8518b5386f86d8044f31fe957307f205541aff92f1637a95bd4c5fff"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-05", "domain": "Forecasting", "id": "mem-03-margin-per-member-order-june", "prompt": "In April, the platform started to comply with the DCWP minimum-pay rule for food delivery workers (Q1 was a tolerated phase-in period), and the new minimum of $19.56 per hour came into force. Whenever a courier's pay for a week falls short of the minimum, the platform tops it up with a true-up. At the same time, our platform updated the dispatcher to restrict when couriers can go online, so the online time we pay for is busier.\n\nBased on what you know and the numbers from January to May (but particularly after the dispatcher change), file a forecast for the per order contribution margin of members' orders in June 2024 (contribution_margin_per_member_order_usd), with an 80% central prediction interval. A member's order is any order placed while the customer held a membership, paid, on trial or partner-sponsored, whether or not the storefront or the basket qualified for a member benefit. We're considering finance's contribution margin per order placed in the month (ORDERED_DATE, every status included), from our platform's order-level P&L as booked. For revenue we keep, we factor in merchant commission, delivery, service and small-order fees, the regulatory response fee, and the membership fee allocated to the order. For costs we bear, we factor in courier pay, the platform-funded share of promotions and refunds, card processing, chargebacks, cancellation costs, referral and quest incentives, and the minimum-pay true-up allocated to the delivery. Two of those terms are Finance's per-order splits of amounts we book at a coarser grain:\n1. The membership fee allocated to the order: each member's fee for a billing period is credited to that member's orders in the period that used a membership benefit (a waived or discounted fee), in the order they were placed. Each order draws up to the value of its own benefit until the period's fee runs out, and later orders get nothing. A cancelled order that was never charged draws nothing.\n2. The minimum-pay true-up allocated to the delivery is the true-up ACCRUED as the delivery settled, not the true-up finally paid. Each time a batch of deliveries settles, the platform sizes what each courier is owed to date for the current payout period (on the hours and pay so far that week); whatever that exceeds the amount already accrued for the courier and period is spread over the courier's deliveries in the batch, in proportion to each delivery's base courier pay. An accrual is never reduced once made, even if the courier catches up later in the week. The accruals are booked daily as XX_ORDERS journals debiting account 5208 (line description MIN_PAY_EXPENSE) and crediting 2110. The payout run later reverses them and books the true-up actually paid (the ADJUSTMENTS line on the COURIER_WEEKLY invoice) in Payables journals; that payment is not a per-order cost.", "version": "b8cb6956cf76f14fb4c87eb78ee220a0ccf40a6081e798acf3afd09db7bf64fd"} {"area": "Growth, Marketing & Membership", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "mkt-06-december-platform-promotion-cost", "prompt": "Marketing's December calendar is set. Alongside the standing offers, Everyday $5 Off and Everyday Free Delivery, both fully funded by us this quarter, and the welcome offer for new customers, we run a Holiday Countdown: free delivery from 29 November to 5 December, cofunded half and half with the restaurants; 20% off from 6 to 15 December, of which we fund 70%; 20% off from 16 to 24 December, of which we fund 30%; a $6 win-back for lapsed Manhattan customers from 15 to 31 December, cofunded half and half; and free delivery from 26 to 31 December, fully ours. The First 3 Orders wave for new customers ended on 22 November and there is no new-customer wave in December.\n\nBased on what you know and the numbers from January to November, file a forecast for the platform-funded promotion cost on orders placed (`platform_promotion_cost_usd`) in December 2024, every status included, with an 80% central prediction interval: our share of every discount redeemed on those orders, the part we book as marketing expense. The restaurants' share of a cofunded discount is theirs, netted from their payouts, and is never our cost.", "version": "9bc614f3148d637e2c5437fb4d6dbded40d25f133ad39c0d4d4d55566e249a8e"} {"area": "Marketplace & Courier Operations", "contracts": {"data_sources": [{"columns": [{"description": "New York calendar month.", "name": "month", "type": "month"}, {"description": "Hours of cases awaiting their first human handling, counted separately.", "name": "waiting_case_hours", "type": "number"}, {"description": "Hours customers had at least one case awaiting first human handling, counting each customer once at an instant.", "name": "waiting_customer_hours", "type": "number"}, {"description": "Customer-hours with at least two cases simultaneously awaiting first human handling.", "name": "multiple_case_customer_hours", "type": "number"}, {"description": "Customer-hours when at least one waiting case had been waiting for seven elapsed days or longer.", "name": "any_week_old_customer_hours", "type": "number"}, {"description": "Customer-hours with waiting cases, all of which had been waiting for seven elapsed days or longer.", "name": "all_week_old_customer_hours", "type": "number"}], "key": ["month"], "max_rows": 5000, "name": "support_customer_waiting_exposure", "order_by": ["month"], "row_count": 3, "title": "Customers waiting for a first human response"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "pe-11-support-customer-exposure", "prompt": "Customer Care wants to know how much of the human refund desk's backlog was borne by the same customers repeatedly. Publish Q3 2024 monthly waiting exposure at both case and customer grain, including simultaneous claims and customers stuck with week-old claims.\n\nA case waits from opening until first human handling starts. Count only time inside the month, including cases opened earlier or handled later. A customer is counted once at an instant in each customer measure, however many orders or cases they have. Seven days means 168 elapsed hours. Report hours to two decimals.\n\nPublish `support_customer_waiting_exposure`, one row per month (ascending): Customers waiting for a first human response\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | New York calendar month. |\n| `waiting_case_hours` | number | Hours of cases awaiting their first human handling, counted separately. |\n| `waiting_customer_hours` | number | Hours customers had at least one case awaiting first human handling, counting each customer once at an instant. |\n| `multiple_case_customer_hours` | number | Customer-hours with at least two cases simultaneously awaiting first human handling. |\n| `any_week_old_customer_hours` | number | Customer-hours when at least one waiting case had been waiting for seven elapsed days or longer. |\n| `all_week_old_customer_hours` | number | Customer-hours with waiting cases, all of which had been waiting for seven elapsed days or longer. |", "version": "deac67e2de416652f594ef44e4a4cb713ccdb116ff494b2403e8b0459195f533"} {"area": "Accounting, Close & Regulatory Reporting", "contracts": {"data_sources": [{"columns": [{"description": "Month the memo created the account-credit liability.", "name": "issue_month", "type": "month"}, {"description": "Whether the credit was created by a refund or by a grant unrelated to a refund.", "name": "origin", "type": "string", "values": ["REFUND", "GRANT"]}, {"description": "Memos creating a positive account-credit liability in this cohort.", "name": "credit_memos", "type": "integer"}, {"description": "Only the spendable account-credit portion of the cohort's memos.", "name": "issued_account_credit_usd", "type": "number"}, {"description": "Account credit applied from these exact memos to purchases by September 30.", "name": "redeemed_by_sep30_usd", "type": "number"}, {"description": "Issued account credit less those redemptions.", "name": "outstanding_sep30_usd", "type": "number"}, {"description": "Sum of these memos' outstanding account-credit balances at each day-end from issuance through September 30.", "name": "outstanding_dollar_days", "type": "number"}], "key": ["issue_month", "origin"], "max_rows": 5000, "name": "account_credit_vintage_runoff", "order_by": ["issue_month", "origin"], "row_count": 12, "title": "Which vintages of account credit were still outstanding?"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "pe-09-credit-vintage-runoff", "prompt": "Finance wants to know how long customer credit stayed unused. Publish the January–June 2024 account-credit vintage runoff through September 30, separating credit created by refunds from other grants.\n\nUse the spendable account credit booked on each memo and its recorded application dates. Credit restored by a later refund belongs to that refund's memo and issuance month. For mixed refunds, include only the account-credit portion. Dollar-days sum outstanding balances at New York day-end from issuance through September 30, inclusive. Include both origins in every month. Report USD and dollar-days to cents.\n\nPublish `account_credit_vintage_runoff`, one row per issue month and origin (ascending): Account-credit issuance, redemption, and balances\n\n| column | type | meaning |\n| --- | --- | --- |\n| `issue_month` | month | Month the memo created the account-credit liability. |\n| `origin` | string | `REFUND` for refund-created credit; `GRANT` for other grants. |\n| `credit_memos` | integer | Memos creating a positive account-credit liability. |\n| `issued_account_credit_usd` | number | Spendable account credit created by these memos. |\n| `redeemed_by_sep30_usd` | number | Credit applied from these memos to purchases by September 30. |\n| `outstanding_sep30_usd` | number | Issued credit less those redemptions. |\n| `outstanding_dollar_days` | number | Sum of these memos' day-end outstanding balances from issuance through September 30. |", "version": "66f9b203e4e190658d5e6031b84c0e2b5b1e264b88f47b7a8660245eec97b8d5"} {"area": "Accounting, Close & Regulatory Reporting", "contracts": {"data_sources": [{"columns": [{"description": "Month the memo created the account-credit liability.", "name": "issue_month", "type": "month"}, {"description": "Whether the credit was created by a refund or by a grant unrelated to a refund.", "name": "origin", "type": "string", "values": ["REFUND", "GRANT"]}, {"description": "Memos creating a positive account-credit liability in this cohort.", "name": "credit_memos", "type": "integer"}, {"description": "Only the spendable account-credit portion of the cohort's memos.", "name": "issued_account_credit_usd", "type": "number"}, {"description": "Account credit applied from these exact memos to purchases by September 30.", "name": "redeemed_by_sep30_usd", "type": "number"}, {"description": "Issued account credit less those redemptions.", "name": "outstanding_sep30_usd", "type": "number"}, {"description": "Sum of these memos' outstanding account-credit balances at each day-end from issuance through September 30.", "name": "outstanding_dollar_days", "type": "number"}], "key": ["issue_month", "origin"], "max_rows": 5000, "name": "account_credit_vintage_runoff", "order_by": ["issue_month", "origin"], "row_count": 12, "title": "Which vintages of account credit were still outstanding?"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "pe-09-credit-vintage-runoff-brief", "prompt": "Publish the January–June 2024 account-credit vintage runoff through September 30, separating refund-created credit from other grants.\n\nUse the spendable account credit booked on each memo and its recorded application dates. Credit restored by a later refund belongs to that refund's memo and issuance month. For mixed refunds, include only the account-credit portion. Dollar-days sum outstanding balances at New York day-end from issuance through September 30, inclusive. Include both origins in every month. Report USD and dollar-days to cents.\n\nPublish `account_credit_vintage_runoff`, one row per issue month and origin (ascending): Account-credit issuance, redemption, and balances\n\n| column | type | meaning |\n| --- | --- | --- |\n| `issue_month` | month | Month the memo created the account-credit liability. |\n| `origin` | string | `REFUND` for refund-created credit; `GRANT` for other grants. |\n| `credit_memos` | integer | Memos creating a positive account-credit liability. |\n| `issued_account_credit_usd` | number | Spendable account credit created by these memos. |\n| `redeemed_by_sep30_usd` | number | Credit applied from these memos to purchases by September 30. |\n| `outstanding_sep30_usd` | number | Issued credit less those redemptions. |\n| `outstanding_dollar_days` | number | Sum of these memos' day-end outstanding balances from issuance through September 30. |", "version": "249b6e4e37529d96cced8b35b55c8b249024136c3d115336da1c70d006d8df60"} {"area": "FP&A & Unit Economics", "contracts": {"data_sources": [{"columns": [{"description": "Month instant cash reached the courier.", "name": "month", "type": "month"}, {"description": "Number of instant cash-outs received.", "name": "cashouts", "type": "integer"}, {"description": "Cash the courier actually received, after the cash-out fee.", "name": "net_cash_received_usd", "type": "number"}, {"description": "Fees for those same cash-outs, counted once.", "name": "fees_usd", "type": "number"}, {"description": "Net cash received multiplied by calendar days from receipt to its pay period's scheduled payout date, summed across cash-outs.", "name": "cash_acceleration_dollar_days", "type": "number"}, {"description": "Cash acceleration dollar-days divided by net cash received.", "name": "cash_weighted_days_early", "type": "number"}, {"description": "365 times total fees divided by cash acceleration dollar-days; a simple decimal rate, not a compounded APR or a mean of individual rates.", "name": "annualized_fee_rate", "type": "number"}], "key": ["month"], "max_rows": 5000, "name": "instant_pay_price_monthly", "order_by": ["month"], "row_count": 3, "title": "The price couriers paid to get paid sooner"}]}, "data_cutoff": "2024-12", "domain": "Dashboards", "id": "pe-07-instant-pay-price", "prompt": "Courier Ops wants to know what couriers paid to get their earnings early. Publish a monthly Q3 2024 view of instant cash received, fees, and how much sooner the money arrived.\n\nGroup by the month cash reached the courier, using New York dates. Compare receipt with the pay period's scheduled weekly payday. Round USD and dollar-days to cents; days and rates to six decimals.\n\nPublish `instant_pay_price_monthly`, one row per month (ascending): The price couriers paid to get paid sooner\n\n| column | type | meaning |\n| --- | --- | --- |\n| `month` | month | Month instant cash reached the courier. |\n| `cashouts` | integer | Instant cash-outs received. |\n| `net_cash_received_usd` | number | Cash received after cash-out fees. |\n| `fees_usd` | number | Fees for those cash-outs, counted once. |\n| `cash_acceleration_dollar_days` | number | Net cash received × calendar days from receipt to scheduled payday, summed across cash-outs. |\n| `cash_weighted_days_early` | number | Cash acceleration dollar-days divided by net cash received. |\n| `annualized_fee_rate` | number | 365 × total fees divided by cash acceleration dollar-days, expressed as a simple decimal rate. |", "version": "9459e4d92a3fffc94ce52814587ad3731b8bdc3013ceaf05dc18a9039b7eec67"} {"area": "Marketplace & Courier Operations", "data_cutoff": "2024-11", "domain": "Forecasting", "id": "fc-20p-quest-desk-vs-finance-read", "prompt": "It is the first week of December 2024. A quest pays a courier a bonus ($15 for 5 deliveries) for working a zone's four-hour window where the marketplace forecast expects more demand than the couriers planning to work it can serve. Mobilisation's November review says the programme, $2.08 million in bonuses, saved about $3.8 million of surge premium and cut no-courier cancellations in the cells it ran by about a fifth. Finance wants an independent read before the Q1 budget.\n\nFile two forecasts for November, each with an 80% central prediction interval: `surge_premium_saved_usd`, what the surge premium in the cells where a quest ran would have come to without the quest, less what it was, summed over those cells; and `no_courier_cancellations_avoided`, the orders cancelled for want of a courier that the quests avoided, the same way. An order belongs to the cell of its pickup zone and the block it was placed in.\n\nFinance's own read of the surge saved is about $1.98 million.\n\nFinance will hold your estimates against that figure and the desk's, so file your own view of November, not theirs.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "db6b1fac8c17a0afbf885f31583a3fdbd218ccdccb069ee260ea746d07fb6576"} {"area": "Trust, Safety & Payments Risk", "data_cutoff": "2024-12", "domain": "Fraud", "id": "rng-30-silent-victims-make-whole", "prompt": "This year 34 storefronts reported that their payouts had been going to a bank account they never set up, switched through the merchant portal; support restored each account and we made the storefronts whole. Payables thinks there are victims who never noticed and are still being paid into the attacker's account.\n\nFind them and work out what we owe each one: every payout run that went to the substituted account, from the switch through the end of the year.\n\nFile one schedule `make_whole`, keyed on `vendor_id`, with one row per storefront still being paid into an account it never set up and `make_whole_usd`, the amount to the cent.", "version": "47d67fa269e25be78db139604af8ef6c2d7542bf46d655739b2b6e3cdbc848ce"} {"area": "FP&A & Unit Economics", "data_cutoff": "2024-10", "domain": "Forecasting", "id": "cb-02-october-orders-lost-disputes", "prompt": "Card disputes on an order keep arriving for weeks after it is placed, and most of what we lose, we lose a month or two later. Finance wants to know what October's orders will have cost us in lost disputes by the time the year closes.\n\nFile one `october_orders_lost_disputes_usd` forecast, with an 80% central prediction interval, for the amounts of disputes on orders placed in October that we will have lost by 31 December.\n\nFinance's own reserve is $182,000: the amount open on October's orders today, grown the way September's matured.\n\nFinance will hold your forecast against that figure, so file your own view, not the reserve.", "prompt_edit": "the entities it names are re-drawn from the public warehouse by the question's own rule (the run drew them from the paper's build)", "version": "ecaa2b4994193a44b64fb50cc78e58aa0e69567f4e1f15b73b85d24ffbefbfb8"}