# What should a workflow do when one step fails?

Canonical URL: https://help.iagenify.com/faqs/workflows/what-should-a-workflow-do-when-one-step-fails

[← Back to Workflows FAQs](https://help.iagenify.com/faqs/workflows)

HELP CENTER · FAQ Workflows 2 min read
# What should a workflow do when one step fails?
Make the failure path explicit: capture the error, preserve enough context to diagnose it, and decide whether to stop, retry safely, or route the item for review. Do not let a later success message hide an earlier failed step.
IaGenify
Workflows

## Separate recoverable errors from terminal ones
A temporary network error may justify a bounded retry, while invalid input or denied access usually requires correction rather than repetition. Classify errors using the information the step provides and set a reasonable retry limit to avoid endless runs.
Keep the input identifier and step name with the failure record so you can locate the affected item. Avoid copying secrets or full sensitive payloads into notifications; include a reference that authorized operators can use to inspect details.
## Make recovery visible and safe
Route unrecoverable cases to a review queue or an alert with a clear next action. If the workflow can partially complete before failing, record which side effects occurred so a recovery run does not repeat them.
Test failures deliberately with a safe sample: invalid data, a temporarily unreachable dependency, and a step that returns no matching result. Confirm that each case ends in the intended state and that the operator can tell what happened. [← Back to Workflows FAQs](https://help.iagenify.com/faqs/workflows)