CSAT is the most widely-used customer experience metric because it is cheap to collect and easy to read. The standard format is a one-question survey: "how would you rate this interaction" on a 1–5 scale, sent after a call or chat.
For AI-handled interactions, the interesting question is the gap between AI CSAT and human-agent CSAT. A good system narrows that gap to within a few points. A great one closes it. A bad one widens it, and you find out only after the customer has churned.
A serious vendor reports CSAT segmented by outcome, calls the agent resolved, calls that handed off, calls that escalated to a supervisor. The blended number alone is not useful.
Vorel reports CSAT per outcome bucket and per vertical, with the raw responses surfaced in your CRM so you can read them yourself.