How it works
Platforms limit request volume to protect their systems, usually per app, per user or per company file, and per second, minute or day. Exceeding the limit returns an error, often HTTP 429 with a header saying when to retry. Some platforms also limit concurrent requests or meter usage in points. Integrations must queue, batch and back off rather than retry immediately.
Example
An invoicing product runs month-end sync for 2,000 customers at once and starts receiving 429 errors. Spreading the jobs and honoring the Retry-After header clears the backlog without losing data.
In integrations
Each platform's limits differ and change. Unify Int is designed to respect per-platform limits, queue requests and return a typed error with a retry hint when a call cannot proceed, so your calls do not fail at month-end. Sync-and-store reads avoid spending the source's limits on every query. Rate limits are a common cause of integration incidents at scale, especially around month-end when every customer syncs at once. Design queues and backoff from day one rather than after the first outage.