ASA, average speed of answer, measures the duration of the wait between the call connecting and the first responsive interaction. A six-second ASA feels fast. A thirty-second ASA feels slow. A two-minute ASA reads as 'they don't care.'
AI receptionists collapse ASA toward zero because there is no queue. The agent picks up immediately on every call. The metric becomes less interesting in steady-state, once you are below five seconds, lowering it further does not move the customer experience, and more interesting at the tail (what is your p95 ASA when call volume spikes).
A vendor that publishes ASA without the percentile breakdown is hiding the tail. A vendor that publishes only the mean is hiding the same thing.
Vorel's ASA is structurally near-zero, every call gets picked up on the second ring. The metric to watch is p95 turn latency, which is published separately.