Engineering notes
Testing escalation: the handoff is part of the product
Most escalation bugs are not about whether the agent escalated, but about what the human received when it did.
An escalation that arrives without context is a worse experience than no automation at all.
Three things to assert
- Trigger correctness, escalate on the defined conditions, not earlier or later.
- Context transfer, order IDs, attempted actions and customer intent carried across.
- Customer communication, the handoff is explained, not a silent transfer.
Over-escalation is also a failure
An agent that escalates everything passes safety checks and destroys your deflection rate. Test both directions.
Scenarios worth building
Abusive language, legal threats, third contact about the same issue, a request outside the tool set, and a customer explicitly asking for a human.
What to take from this
- Assert the payload of the handoff, not just that it happened.
- Over-escalation deserves a test too.
- Escalation quality decides whether deflection is real.
See it on your own agent
Connect an endpoint and run twenty scenarios. The first run usually finds something.
Start testingKeep reading
- Claimed, not executed
The most expensive agent failure is a confident sentence about something that never happened, and a transcript cannot see it.
- Not verified is not a pass
A test suite that quietly passes the checks it could not perform is worse than no suite at all.