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: Trustpilot providerId: trustpilot created: '2026-05-04' # Provenance stamped 2026-08-11: this artifact was written by the API Evangelist # bulk sweep dated 2026-05-04, not harvested from the provider. See roadmap#35. method: generated modified: '2026-05-05' reconciled: true tags: - Reviews - Reputation - Rate Limiting description: Trustpilot publishes recommended rate-limit ceilings rather than per-second numerics. The guidance is to stay under 833 calls per 5 minutes and 10,000 calls per hour, and to architect around webhooks, token reuse, and server-side caching to reduce call volume. API access itself is bundled with Premium / Enterprise plans as an add-on. sources: - https://developers.trustpilot.com/rate-limiting - https://developers.trustpilot.com/ limits: - name: 5-Minute Soft Ceiling scope: api-key metric: requests_per_5_minutes limit: 833 timeFrame: minute notes: Trustpilot recommends not exceeding 833 calls per 5 minutes (approx. 167/minute). - name: Hourly Soft Ceiling scope: api-key metric: requests_per_hour limit: 10000 timeFrame: hour notes: Trustpilot recommends a maximum of 10,000 calls per hour. policies: - name: Webhook-First Architecture description: Trustpilot recommends consuming review and event data via webhooks instead of polling to avoid breaching the 5-minute and hourly recommendations. - name: Token Reuse description: Reuse OAuth 2.0 access tokens across requests rather than minting a new token for every call; minting incurs additional API calls that count toward the limits. - name: Server-Side Caching description: Cache review data on your backend rather than fetching directly from frontend page views; this is the primary lever for staying under the published ceilings. maintainers: - FN: Kin Lane email: kin@apievangelist.com