# Lead-routing recovery runbook

Template for the team operating the workflow. Adapt before a pilot.
The sample payloads are fictional; they do not connect to a real CRM.

## Name the owners
- Workflow owner:
- CRM connection owner:
- Fallback sales queue owner:
- Escalation route and response window:
- Location of retained submissions and delivery state:

## Investigate
1. Find the stable submission ID and the intended destination.
2. Record the execution ID, timestamp, last completed step and failed step.
3. Inspect the CRM for the operation/record ID before attempting a new write.
4. Inspect notification delivery separately; a saved lead does not prove a message was sent.
5. Classify the issue: invalid input, routing gap, authentication, temporary rate limit, timeout or partial success.

## Recover
- Invalid input: retain context and ask the review owner to repair or reject it.
- Routing gap: assign the accountable fallback queue; record the manual decision.
- Authentication: connection owner repairs access; replay retained submissions under duplicate protection.
- Rate limit: respect the service guidance; use bounded attempts and elapsed time; escalate when exhausted.
- Timeout: reconcile whether the destination committed the operation before retrying.
- Partial success: resume the failed side effect while retaining the completed write.
- Duplicate: use a durable atomic guard / supported idempotent operation. A read-then-write check alone can race.

## Verify and close
Confirm the intended CRM record, owner, single required message and final delivery state.
Link the evidence, note any manual repair and tell the appropriate stakeholder the outcome.
Review repeated failures, task/execution usage and retained-submission age during the pilot.
Test any changed mapping or retry policy in a sandbox before using it in the live process.

Reference guide: https://www.derrick.dk/post/n8n-vs-zapier-lead-routing