Reliable operations
Plan for retries.
Avoid duplicate work.
Connection failures and repeated deliveries are part of integration work. The key is to prevent them from creating duplicate operations.
A practical example
Your application submitted a booking, but the response was lost. Before creating another operation, it needs a way to recognise the original request.
Keep business references stable
When creating a booking, our REST API recognises an existing externalReference from the same origin and returns the existing booking. Use a stable reference for the same operation. This mechanism does not update an existing booking and does not automatically apply to every API operation.
Distinguish events from deliveries
A webhook redelivery receives a new delivery ID. Our recommendation: recognise processed events by their event ID instead of relying only on the delivery ID. This prevents the same change from triggering the same follow-up action repeatedly.
Handle errors deliberately
Our REST API returns structured errors with an error code and traceId. Use them for diagnostics and support without logging credentials. Correct invalid input or missing permissions before retrying. Space out retries for temporary failures and set a clear limit.