# What is Escalation Path?

Also called escalation route.

An escalation path is the predefined route by which work moves from an automated process to a person, specifying the trigger conditions, the recipient, the context transferred, and the fallback when the recipient is unavailable. It is the mechanism that keeps a case from stalling silently when a system reaches the limit of what it should decide.

A complete path defines four things. Triggers state the conditions that force a handoff, whether a rule, a confidence threshold, a keyword, or an explicit customer request. The recipient is a named role. The transferred context is what the person receives along with the case. The fallback names what happens when the recipient does not respond within a stated window.

The fallback is the part most often left undefined and the part that fails. A path routing to one person with no timeout means every absence becomes a stalled case, discovered later by the customer. A minimum viable arrangement is a named backup and a time limit after which the case surfaces somewhere visible rather than sitting in an individual queue.

Context transfer quality determines whether escalation helps. Handing over a bare ticket identifier forces the recipient to reconstruct everything, which is slow and irritating for the customer who has to repeat themselves. A useful handoff carries the conversation, what the agent already attempted, the specific reason for stopping, and the records already retrieved.

Certain categories should escalate on first contact rather than after an automated attempt fails. Complaints, legal or regulatory matters, safety concerns, distress, and anything involving a dispute or a cancellation generally belong to a person immediately. Attempting these automatically and escalating afterward produces a worse outcome than routing them directly, because the failed attempt itself becomes part of the grievance.

## Key points

- Defines triggers, recipient, transferred context, and fallback
- Undefined fallback is the most common failure point
- Handoff must carry attempts made and reason for stopping
- Complaints and disputes escalate immediately, not after a failed attempt
- Unanswered escalations must surface, not sit in one queue

## In practice

A booking agent escalates when a guest mentions a refund, when a reservation spans a blocked date, or when the guest asks for a person. The case moves to the duty manager with the full conversation, the reservation record, and a one line reason. If untouched after thirty minutes it posts to the operations channel where any manager can pick it up.

## Related terms

- [Exception Handling](/en/glossary/exception-handling)
- [Ticket Deflection](/en/glossary/ticket-deflection)
- [First Contact Resolution](/en/glossary/first-contact-resolution)
- [SLA For Automated Work](/en/glossary/sla-for-automated-work)
- [Scope Of Work](/en/glossary/scope-of-work)

[Back to the AI Glossary](/en/glossary)