Agent Escalation
What is Agent Escalation?
When required, such as for complex or atypical requests, self-service interactions like those with chat bots can be escalated to an agent with full context and conversation transcript.
Agent escalation is the point where an automated or self-service conversation is passed to a human agent. It usually happens when a request is complex, unusual, emotionally sensitive or outside what a bot is allowed to handle, or when the customer simply asks for a person. A good escalation carries the conversation transcript, the customer’s identity and the reason for the transfer, so the agent can pick up where the bot left off.
Escalation design matters because a poor hand-off forces customers to repeat themselves, which lengthens handle time and erodes trust in self-service. Teams typically track escalation rate, the reasons behind escalations and whether escalated contacts are resolved on the first attempt. A high escalation rate for one intent often points to a gap in the bot’s knowledge or a broken process upstream.
Related: Human-in-the-loop, AI agents, Self-service, UWXP.