Rate limits and versioning
How the Adnova APIs are rate-limited and how we version breaking changes.
Rate limits
Manage API (/manage/v1) and Analytics API (/analytics/v1)
Both surfaces use the same policy in each service:
-
100 requests per 60 seconds per authenticated user. The count applies to all routes on that API (not per endpoint).
-
There is no separate burst pool beyond that per-user quota in the active window on each service.
-
Response headers use the common
RateLimit*-style fields (IETF rate-limit header draft) so you can read remaining quota and when the window resets. Check those headers on the response; 2xx JSON bodies do not repeat that information. -
When exceeded, the API returns 429 with a JSON body like below. For this case, rely on the error payload, and any
RateLimit*headers on the response to see when you can try again.{ "error": { "code": "RATE_LIMIT_EXCEEDED", "message": "Rate limit exceeded" } }
Analytics refresh (POST /analytics/v1/refresh)
In addition to the global limit above, refresh uses a separate cooldown so expensive platform work cannot be spammed:
- One successful refresh per
(platform, adAccountId, datePreset)every 15 minutes. - If you hit this limit, the response is 429 with a
Retry-Afterheader (seconds) and a JSON body that includesretryAfterSeconds.
Ad Launcher API (/ad-launcher/v1)
Ad Launcher splits read and write into two independent buckets:
- General —
~100 req / 60 sper user. Applies to lookups, the creative-upload handshake, andGET /launches/{batchId}. - Launch (write) —
POST /launchesonly. Much stricter; quota is configured per org inadnova_ad_launcher.
Because the buckets use distinct Redis key prefixes, spamming POST /launches does not eat into your read budget. See Ad Launcher rate limits for details.
Versioning strategy
The API is versioned in the URL path — e.g. /manage/v1/..., /analytics/v1/..., /ad-launcher/v1/.... A new major version only ships when we need to make a breaking change.
