--- name: win-loss-recap-writer description: Writes a crisp win/loss recap when a deal closes by pulling its history, stage path, and close reason into a short lessons-learned the team can act on. Use when a deal is marked closed-won or closed-lost, or the user asks to "write a win/loss recap", "what happened with this deal", "lessons from this close", "why did we lose this". license: MIT compatibility: Requires a connected CRM. Read-only. metadata: author: Builder.org version: 2.0.0 persona: Strategic Steve rave: R4 A3 V3 E3 category: crm --- # Win/Loss Recap Writer ## What it does When a deal closes, either way, this writes the short honest recap: what happened, why it really closed the way it did, and the one or two lessons worth carrying into the next deal. Wins get repeated and losses get learned from only when someone writes this down within a week, while the evidence is still in the CRM and the rep still remembers. Read-only; it reports, it changes nothing. ## When to use - "Write the win/loss recap for [deal]", "what happened with this deal", "lessons from this close", "why did we lose this". - Triggered when a deal moves to closed-won or closed-lost. - Not for reviving deals that are stalled but still open; that is `stalled-deal-revival`. This skill starts when the deal ends. ## Inputs A closed deal, won or lost, from any CRM your assistant can read (HubSpot, Salesforce, or similar). Optional: the rep's quick take, which the recap quotes and tests against the record rather than simply adopting. ## The bar - One primary driver, with evidence. "Price, and also timing, and also the champion left" is a recap dodging its job; pick the driver the record best supports and file the rest as contributing. - Every claim traces to the record: an engagement, a stage timestamp, a recorded reason, or the rep's take, cited as such. An inference is labeled an inference. - Lessons generalize. "Lakeview wanted more training" is an observation; "deals under $15k stall when onboarding looks heavier than the price" is a lesson. - Short enough to be read: one screen. The recap nobody reads teaches nobody anything. ## Workflow 1. **Pull the record.** Stage path with time in each stage, amount, cycle length, competitors if noted, the recorded close reason, and the engagement history around each stage change. 2. **Reconstruct the arc.** The three or four turning points: where momentum appeared, where it stalled, what re-opened it or ended it. Stage timestamps plus surrounding engagements tell this story more honestly than memory does. 3. **Name the driver.** The primary reason it closed the way it did, backed by the strongest evidence in the record. Where the recorded close reason and the engagement evidence disagree, say so; that disagreement is often the most valuable finding in the recap. 4. **Test the rep's take** when given: quote it, then note where the record confirms or complicates it. Respectfully; the rep knows things the CRM does not, and the CRM remembers things the rep does not. 5. **Extract one or two lessons** that would change behavior on similar deals: a stage to slow down in, a stakeholder to multi-thread earlier, a discount pattern that did not buy anything. Tie each to the evidence. 6. **Write the recap** in the output format, one screen, no blame in either direction. ## Output ``` ## [Deal] · [WON/LOST] recap **Size:** $[amount] · **Cycle:** [days] · **Stages:** [path with days in each] · **Competitor:** [if any] ### What happened [2-4 sentences: the arc, told by the record] ### Why we [won/lost] [Primary driver + evidence. Contributing factors in one line. Inference labeled as inference.] ### Rep's take (if given) "[quote]" · [where the record agrees or complicates it] ### Lessons 1. [Generalizable lesson tied to evidence] 2. [Optional second] ``` ## Worked example ``` Fictional example. Never reuse these names or numbers in real output. ## Brightpath Staffing onboarding suite · LOST recap **Size:** $44k · **Cycle:** 94 days · **Stages:** Discovery 12d, Demo 9d, Proposal 61d, Closed Lost · **Competitor:** none recorded ### What happened Fast, warm start: discovery to demo in three weeks, with the ops director driving. Then the proposal sat for two months. Engagement dropped from weekly to nothing within ten days of pricing going over. Two follow-ups went unanswered; the recorded close reason is "went with internal solution." ### Why we lost Primary driver: the proposal priced the full rollout while the record shows appetite for a starting point; Dana's last substantive email asked "is there a way to phase this?" and the reply restated the full package. The recorded reason ("internal solution") is what they built in the gap we left. Contributing: single-threaded throughout; no second contact ever engaged. The driver is an inference from the engagement record; the close reason field alone would have hidden it. ### Rep's take "They never had budget." · The record complicates this: they asked about phasing, which is what buyers with partial budget do. Budget shaped the loss; the unphased proposal decided it. ### Lessons 1. When a buyer asks to phase, the next artifact is a phased proposal, not a restatement. This deal's 61 silent days started the week we skipped that. 2. No deal over $25k should reach Proposal single-threaded; there was no second voice to keep it alive when Dana went quiet. ``` ## Never - Never accept the close-reason field as the whole truth when the engagement record tells a richer story; report both. - Never dress an inference as a fact. "The record suggests" is the honest register. - Never write a blame piece, in either direction. The unit of analysis is the deal, not the rep. - Never list more than two lessons. A recap with six lessons has none. ## Recovery - Empty close reason and thin engagements: reconstruct what the timestamps alone support, label the driver low-confidence, and let the first lesson be about record hygiene, stated kindly. - Won but nobody knows why: the record usually does; look for the artifact or moment right before acceleration and name it as the probable driver, labeled as inference. - Rep's take contradicts the record outright: present both without arbitration and flag the deal for a human conversation; the recap's job is honesty, not adjudication. - The deal closed months ago: write it anyway, note the staleness, and weight the written record over anyone's memory, including the rep's.