generated: '2026-08-13'
method: searched
source: >-
  https://www.refersion.dev/reference/welcome-to-refersion, the provider-published OpenAPI harvested
  to openapi/_original/refersion-rest-api-readme-harvest.json, and https://www.refersion.com/pricing/
name: Refersion Rate Limits
description: >-
  Rate-limiting posture for the Refersion REST and GraphQL APIs. Refersion publishes NO
  request-rate limit — no requests-per-second, no requests-per-minute, no daily quota — for any
  surface, and returns NO rate-limit response headers. What the contract DOES declare is a set of
  hard payload-size caps, and it signals one of them with HTTP 429, which is the single most
  misleading thing about this API's throttling behaviour: a 429 from Refersion means "your array is
  too big", not "slow down", so an exponential-backoff retry of the identical request will fail
  forever.
documented_request_rate_limits: 0
limit_count: 5
response_headers:
  x_ratelimit: false
  ratelimit_rfc9331: false
  retry_after: false
  note: >-
    No X-RateLimit-*, RateLimit-* or Retry-After header is documented on any operation, and none is
    declared in the published OpenAPI. An agent has no runtime signal to pace itself by.
exhaustion_status: null
limits:
- name: Conversion triggers per delete request
  scope: per-request
  type: payload-size
  limit: 50
  unit: elements
  operations:
  - delete_conversion_trigger
  status_on_exceed: 429
  body_on_exceed: '{"error": "Max number of elements on attribute (50) reached."}'
  evidence: Declared as the 429 response on POST /trigger/delete in Refersion's own OpenAPI.
  remediation: Split into batches of 50 or fewer. Retrying unchanged will not succeed.
- name: SKU records per commission request
  scope: per-request
  type: payload-size
  limit: 50
  unit: skus
  operations:
  - new_sku_commission
  status_on_exceed: 422
  body_on_exceed: '{"errors": ["Only 50 sku records can be processed per request."]}'
  evidence: >-
    Declared as a 422 message on POST /offer/new_sku_commission, and as "Max: 50" on the skus field.
- name: Affiliate IDs per status change
  scope: per-request
  type: payload-size
  limit: 50
  unit: ids
  operations:
  - affiliate_status_change
  status_on_exceed: unknown
  evidence: >-
    "Limit of 50 IDs per call" — stated verbatim in the operation description published in
    https://www.refersion.dev/llms.txt and on the reference page.
- name: Affiliate list page size
  scope: per-request
  type: pagination
  limit: 100
  unit: records
  operations:
  - list_affiliates
  evidence: '"Total that should be returned per API call. Maximum of 100 per call." — the `limit` field description in the published spec.'
- name: Report download link validity
  scope: per-resource
  type: ttl
  limit: 120
  unit: seconds
  operations:
  - get_reporting_link
  evidence: >-
    "Links expire in 2 minutes from the time of request for security purposes." — operation
    description in llms.txt and the reference page. Response returns `expire_time` as a Unix epoch
    integer.
surfaces:
- surface: REST API v2
  base_url: https://api.refersion.com/v2
  request_rate_limit: undocumented
- surface: GraphQL API
  base_url: https://api.refersion.com/graphql
  request_rate_limit: undocumented
  note: >-
    No published query-complexity or depth limit. The endpoint refuses anonymous introspection
    (HTTP 403 on 2026-08-13), so no limit could be observed either.
- surface: Inbound order tracking webhook
  base_url: https://inbound-webhooks.refersion.com
  request_rate_limit: undocumented
- surface: Outbound webhooks
  request_rate_limit: undocumented
  note: No documented delivery-rate cap, retry schedule or backoff policy. See asyncapi/refersion-webhooks.yml.
plan_gating:
  note: >-
    Access to the developer API is a plan boundary rather than a rate boundary — the pricing page
    lists API access starting at the Growth tier ($199/mo, or $159/mo billed annually). Plan-gated
    features fail with 422 "Your account plan does not support this feature (error 1)." rather than
    402 or 403.
  source: https://www.refersion.com/pricing/
gaps:
- No published request-rate limit for any surface.
- No rate-limit response headers, so clients cannot self-pace.
- >-
  HTTP 429 is overloaded to mean a payload-size violation, which will mislead any generic client
  library that implements standard 429 backoff.