Announced: 04.11.2026
Effective date: 04.11.2026
Affected endpoints: /v2/compensations and /v2/compesations/types

What's changing

To improve the overall stability and security of our APIs, we are introducing the following changes to the compensation APIs.

1. Limit on parallel requests without a person.id filter

We are introducing a concurrency limit for requests without a person.id filter: each company can have at most 2
such requests in progress at the same time. Requests over the limit are rejected with:

• 429 Too Many Requests, header Retry-After: 30, error type compensations.too_many_requests

Requests filtered by person.id are not counted towards this limit.

This is separate from the existing rate limit, which also returns 429 but with no Retry-After header and no response body; wait 1 second before retrying those.

2. Time limit on a single request

A request that takes too long to process is stopped after a time limit and rejected with:

• 422 Unprocessable Content, error type compensations.unprocessable_content, no Retry-After header

Sending the same request again will not help, because the time depends on the amount of data requested, not on the current load.

3. Temporary unavailability

If the Compensations API is temporarily unavailable, requests are rejected with:

• 503 Service Unavailable, header Retry-After: 30, error type compensations.service_unavailable

What you need to do

• On 429 or 503: wait for the number of seconds in the Retry-After header (30) before retrying. Stop after a few attempts instead of retrying indefinitely.
• On 422: do not retry the same request. Filter by person.id instead (for example
  person.id=100,200,300); if the request is already filtered, send fewer IDs per request.
• Set your HTTP client timeout to at least 60 seconds, so your integration receives the 422 instead of timing out on its own side.
• Where possible, filter by person.id. These requests return faster and are not affected by the parallel request limit.

If you need additional support, please contact us via your Partner Contact or Support.