Because nobody baselined. The assistant goes live, everyone agrees it feels faster, and six months later the renewal conversation has no evidence in it either direction.
Pick three or four things you can count now: median time to first reply, how often a question is escalated, how long a standard account question takes to answer, ticket volume by cause.
Measure them for a month before enabling anything. This is the step everyone skips and the only one that makes the rest meaningful.
Time to first reply, escalation rate, volume by cause — before enabling.
Only alongside whether the customer came back.
After the novelty period — a month or two in.
Half an hour with your own data usually saves reading three of these. The guides will still be here afterwards.