Containment measures the bot. Who is measuring the customer?

When a bot handles a conversation from start to finish, it counts as contained. The customer never reached an agent. On a dashboard that looks like success, and in a business case it looks like saved cost.

The trouble is what containment cannot see. A customer who gets the answer they needed is contained. So is a customer who gives up, closes the chat and rings back tomorrow. So is the one who goes to a competitor. The metric scores all three the same.

A customer who gives up looks exactly like a customer who got what they needed.

We have been here before

First contact resolution went through the same thing. For years it was measured from the system side, by whether a case was closed or whether a repeat call arrived within a set window. Then contact centres started asking customers whether their issue was resolved, and the two numbers rarely agreed.

The customer's answer won, because it was the one that predicted whether they stayed.

Ask the customer

The fix is the one that worked for first contact resolution. Ask the customer, on the automated journey, straight after it ends. Did you get what you came for? How easy was it? Would you use it again?

Put that answer next to the containment figure and you have a number a board can trust. You can see where the bot resolves issues and where it only ends conversations. You can move work to automation where it holds up, and keep people where they matter.

Your AI investment deserves a measure that tells the truth about it.