specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Awin
providerId: awin-affiliate
created: '2026-07-05'
modified: '2026-08-13'
reconciled: true
generated: '2026-08-13'
method: searched
source: https://help.awin.com/apidocs/introduction-1
limit_count: 6
tags:
- Affiliate Marketing
- Advertising
- Rate Limiting
- Quotas
- Throttling
description: >-
  Awin operates a single platform-wide throttle across its API. The documented
  limit is 20 API calls per minute per user; because tokens are set at the user
  level, the limit is shared across every publisher and advertiser account the
  token can reach. Some endpoints impose their own request-scoping constraints
  rather than a separate rate limit - for example, the transactions endpoints
  cap each request to a date range of at most 31 days.
notes: >-
  The 20 calls-per-minute figure is documented in the Awin API introduction.
  Confirm the current value on the introduction page during reconciliation.
sources:
- https://help.awin.com/apidocs/introduction-1
- https://help.awin.com/apidocs/returns-a-list-of-transactions-for-a-given-publisher
responseCodes:
  throttled: 429
responseHeaders:
  documented: false
  headers: []
  retryAfter: false
  notes: >-
    Awin publishes the NUMBER but no runtime signal. No X-RateLimit-*, no
    RateLimit-* (RFC 9239 draft), no Retry-After is documented anywhere in the
    API reference, and none was observed on live unauthenticated responses from
    api.awin.com on 2026-08-13. An agent cannot discover its remaining budget and
    must count its own calls client-side. This is the single most consequential
    runtime gap in the Awin API for automated consumers.
limits:
- name: Requests Per Minute
  scope: user
  metric: requests
  limit: 20
  window: 60s
  notes: >-
    Platform-wide throttle shared across all accounts the user token can access.
    Documented on the API introduction page.
  source: https://help.awin.com/apidocs/introduction-1
- name: Create Offers Requests Per Minute
  scope: endpoint
  metric: requests
  limit: 50
  window: 60s
  notes: >-
    The Post Offers page states a 50 calls-per-minute limitation on that surface,
    which contradicts the platform-wide 20/min figure on the introduction page.
    Awin does not reconcile the two. Recorded as published; treat 20/min as the
    safe planning figure.
  source: https://help.awin.com/apidocs/offers-api
- name: Transaction Date Range
  scope: request
  metric: days
  limit: 31
  notes: Maximum span between startDate and endDate on the transaction list endpoints.
  source: https://help.awin.com/apidocs/returns-a-list-of-transactions-for-a-given-publisher
- name: Conversion API Batch Size
  scope: request
  metric: orders
  limit: 1000
  notes: >-
    Maximum orders per batch on POST /s2s/advertiser/{advertiser_id}/orders. The
    docs also prescribe a batching CADENCE - not shorter than 10 seconds (to stay
    inside the call limit) and not longer than 60 seconds (or the S2S call loses
    tracking-method priority).
  source: https://help.awin.com/apidocs/conversion-api
- name: Batch Validation Size
  scope: request
  metric: transaction objects
  limit: 40000
  notes: Maximum transaction objects in one POST /advertisers/{advertiserId}/transactions/batch request.
  source: https://help.awin.com/apidocs/approve-decline-amend-batch-transactions-for-a-given-advertiser
- name: Link Builder Short-Link Quota
  scope: account
  metric: short links
  limit: null
  window: 24h
  notes: >-
    A per-publisher 24-hour cap on generated short links. The value is
    account-specific and not published - it is readable at runtime from GET
    /publishers/{publisherId}/linkbuilder/quota, which returns limit (max in
    24hrs) and usage (generated in the last 24hrs). This is the ONLY runtime
    quota signal Awin exposes anywhere in the API.
  source: https://help.awin.com/apidocs/quota
policies:
- name: User-Level Scope
  description: >-
    Rate limiting is applied per user, not per account. A token that reaches
    many accounts still shares one 20-call-per-minute budget.
- name: Backoff Strategy
  description: >-
    Clients should throttle to stay under 20 calls/min, implement exponential
    backoff with jitter on 429 responses, and batch date ranges in chunks of up
    to 31 days for transaction pulls.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com