Abandonment rate

The two senses share a name and little else. Outbound abandonment happens when a dialler places more calls than it has capacity to handle, so an answered call has nobody to connect to. Inbound abandonment happens when the caller gives up in a queue. The first is caused by over-dialling, the second by under-staffing, and they are measured and fixed differently.

The outbound sense is regulated because the recipient experience is a silent or dead call. Rules in several jurisdictions cap the permitted abandonment percentage over a defined measurement period and require an informative recorded message within a short window when a call is abandoned. Predictive diallers must therefore pace against the cap, not merely against available capacity.

AI agents change the calculus in one direction and add a new failure mode in the other. Because an agent is available the instant a call connects, classic outbound abandonment from over-dialling largely disappears. But a new equivalent appears: calls where the agent connects and then fails — a provider outage, an unhandled request, a caller who hangs up on realising it is automated. Tracking that as its own category is more useful than folding it into a legacy metric.