Skip to main content
Some Salesforge and Multichannel API endpoints limit how fast each account can call them, so that one integration cannot slow down everyone else. A request over a limit gets 429 Too Many Requests and a Retry-After header that says how many seconds to wait.

How the limits work

Limits apply to your account, not to an API key or an IP address. Every API key and integration on the account counts toward the same limits. Each limit applies to one endpoint and has three parts:
  • Requests per minute is the pace you can keep up.
  • Burst is how many requests you can send at once before you have to slow down to that pace. Unused requests build back up to the burst at the per-minute rate. For example, 60 requests per minute with a burst of 10 lets you send 10 requests at once, then 1 a second.
  • Requests in flight is how many requests can be running at the same time.

Rate limited endpoints

These endpoints are rate limited: Salesforge API paths are relative to https://api.salesforge.ai/public/v2. Multichannel API paths are relative to https://multichannel-api.salesforge.ai/public/multichannel. Support can set different limits for your account.

When you hit a limit

A request over a limit gets 429 Too Many Requests. The body says how long to wait:
When too many of your requests are running at the same time, it reads too many concurrent requests for this endpoint, retry after 1 second instead. The Retry-After header carries the same wait, in whole seconds. It is always at least 1. A rejected request is not processed, so you can send it again unchanged.

Rate limit headers

Responses from a rate limited endpoint carry three headers that describe its limit, so you can pace yourself before you hit it: A 429 for too many requests in flight carries only Retry-After.

Retrying well

  • Wait for Retry-After before you send the request again. Don’t retry in a tight loop.
  • Slow down when RateLimit-Remaining gets close to 0.
  • Keep the number of requests you run in parallel below the requests in flight limit. Running more at once doesn’t make you faster; the extra requests get 429.
  • Read lists with the largest page you need, and read the pages one after another. For contacts, use cursor pagination. See Getting paginated contacts.
  • Don’t poll for changes by reading a whole list again and again. React to webhooks where they cover your case.
This example waits for Retry-After and sends the request again: