Systems that answer
How often a person takes over — and whether they arrive holding the conversation or starting one.
Also called Handover rate · Доля эскалаций
Formula
Escalation rate = Conversations passed to a person ÷ Conversations
- Passed to a person
- a human took over, whether the system decided or the customer asked
- Conversations
- same denominator, same window
Worth splitting by who initiated it. A system escalating because retrieval found nothing is working; a customer demanding a human after three unhelpful replies is the same number describing the opposite situation.
The rate on its own says almost nothing — what it costs depends entirely on the state the person inherits. An escalation that arrives with the history, the qualification state and the open question is a warm conversation somebody continues. An escalation that arrives as a fresh chat window is the customer explaining themselves twice, and it is the point at which most deployments lose the sale.
So the number to report beside it is not the rate but the restart: how many escalations required the customer to repeat something they had already said. That figure is knowable from the transcripts and is almost never collected.
A deployment reporting a falling escalation rate and a flat conversion rate has moved conversations out of the escalation queue and into nothing. Before optimising the rate, measure what the handover currently costs — whether the person receives the thread, and whether the customer has to start again. Fixing that usually improves the outcome more than removing escalations would.
Quantities derived from the counts. Each is only as sound as the definitions beneath it.
Cannot be computed without
- Conversation
- One exchange between a person and the system, from their first message to the point it is over — and where that point is decides every rate computed below it.
Knowing the definition is not the same as being able to check the figure. These are the procedures that do the second thing.
- Accepting an automated agent
- “The agent is trained on your data and ready to go live.” · 45 minutes, 6 questions.
The definitions are the easy part. Whether the figure on your dashboard was computed this way is a different question, and usually the more expensive one.