When an n8n workflow times out, the timeout is usually the symptom.
The cause is often somewhere else: slow upstream APIs, serialized work that should be batched, or unbounded retries that turn one problem into fifty.
Start with shape
Before changing timeout settings, map the workflow:
- Which node waits longest?
- Which API is least predictable?
- Which loop grows with customer volume?
Fix that shape first.