Glossary
Call Deflection
The share of inbound calls resolved without ever reaching a human agent — typically expressed as a percentage of total volume.
Call deflection measures how many inbound calls are fully resolved by automation (AI voice agent, self-serve flow, or knowledge-base answer) without consuming a human-agent minute. For example: 1,000 inbound calls, 600 fully resolved by AI, 400 escalated to humans = 60% deflection rate. Industry benchmarks for AI voice agents in well-tuned deployments typically land at 35–55% deflection within 30 days and 50–70% within 90 days, with the exact number depending heavily on call-mix complexity. The metric only counts genuine resolution — a call that "deflects" but the customer calls back unresolved later doesn’t count.
Why it matters
- Direct cost reduction: every deflected call saves 5–12 minutes of human agent time.
- Improves response time on remaining human-handled calls (shorter queues).
- Reallocates agent capacity to higher-value work: tier-2 troubleshooting, retention, outbound.
- Reduces hiring pressure during seasonal volume spikes.
- Differentiates "true deflection" from "deferral" — a call that hangs up isn’t deflected.