When a bot should give up, and how to transfer without making the customer start over.
Every automated support system needs an exit to a person. How that exit is designed determines whether customers experience the bot as helpful or as an obstacle.
Know when to stop trying
Define explicit triggers: the customer asks for a person, two consecutive failures to understand, detectable frustration, or any topic on a defined escalation list such as complaints, cancellations and anything with a legal or safety dimension. A bot that keeps trying past those points is doing damage rather than deflection.
Carry everything across
The agent should receive the full transcript, the customer's identity and history, what the bot understood the problem to be, and what it already tried. Asking the customer to repeat themselves after a five-minute conversation is the moment goodwill is lost, and it is entirely avoidable.
Be honest about the wait
Say whether a person is available now, give a realistic wait, and offer a callback or email if they are not. Silently queueing someone who was told they would be connected is worse than telling them the office is closed.
Measure post-handoff resolution rate and customer satisfaction separately from bot containment. Optimising containment alone reliably produces a system customers resent.
Want this for your business?
Let's talk about how we can help you build and grow.
