Viable Call Success: the headline metric, and what sits beneath it

All customer calls, Jul 8 – Aug 7 2026. Of calls with genuine booking intent, our AI handled 96.4% successfully — the like-for-like number to put beside competitor claims. This report defines that metric precisely, shows the stricter metrics underneath it, and details what each one would take to improve.

1One month, three honest metrics

These are the same 15,433 viable calls measured three ways, from most to least generous. Each is truthful; they answer different questions. When a competitor quotes “80–90%,” it is almost certainly a success/containment metric — the top bar — not a booked-jobs rate. Compare like with like.

Guardrail for marketing: quote Viable Call Success with its definition (“% of booking-intent calls the AI handled successfully”) and never place it in the same sentence as a competitor’s number without confirming what their denominator and numerator are. The stricter metrics below exist so we never have to defend a number we can’t reproduce.

2Where inbound calls go

Just over a third of inbound production calls carry booking intent (“viable”), and just under a quarter of those end in a CRM booking during the call. The rest of this report works through that middle segment.

3What drags success below 100%

Only 556 viable calls were classified unsuccessful. This is the improvement backlog for the headline metric — small, concrete, and dominated by drop-offs during information gathering, failed transfers, and callers who wouldn’t talk to an AI.

4What “viable” currently includes

Viable is a per-call LLM judgment (“the call was related to booking an appointment”), made independently of the call-end category. Several categories that are marked viable convert at effectively zero — context for why the booking-rate metrics below look the way they do.

5The booking-rate generosity ladder

If we also want the internal booking rate to be more defensible, each step removes one arguable sub-category from the viable denominator, cumulatively. Booked jobs (the numerator) are unchanged. This tops out around 31% — useful honesty: no denominator surgery reaches success-metric territory.

6The measurement gap: bookings we can’t see

The booking numerator only counts crmJobBooked / crmJobRescheduled. Accounts without a working CRM integration can never contribute a booking — but their viable calls still count against the aggregate.

7Daily volume over the window

8The genuine conversion gap

After every defensible exclusion, these viable-but-unbooked calls remain. A success-style headline doesn’t make these disappear — they are the revenue lever, and the reason to keep the strict booking rate alive internally.

9Per-customer view

Customers with ≥50 viable calls in the window: the success rate alongside today’s booking rate and the booking rate under the generous definition. Rows marked “no CRM sync” recorded zero CRM bookings all month — their true booking rate is unknown, but their success rate is still measured.

10Recommendations

  1. Adopt Viable Call Success (96.4%) as the marketing headline. Publish it with its definition — “percentage of booking-intent calls the AI handled successfully” — and its companion stat (“…and 41.8% ended booked, transferred, or fully resolved on the call”). The definition is what makes the number safe: it is reproducible from our own postcall classifications, and it is the honest like-for-like against competitor “containment” or “success” claims.
  2. Work the 556-call failure backlog to push the headline higher. Information-gathering drop-offs (160), failed transfers (100), and AI refusals (93) are the top three. Each is a product fix, not a definition fix — and at this base, resolving ~230 failures moves the headline from 96.4% to ~98%.
  3. Fix booking measurement so the internal metric stops lying. 44 accounts with ≥20 viable calls recorded zero CRM bookings in 30 days (4,326 viable calls — 28% of the denominator, structurally unable to contribute a booking), and 718 further viable calls were classified booked by the postcall LLM with no CRM job recorded. Repair integrations or fall back to CallResult = "Booked" for non-integrated accounts.
  4. Align the Viable flag with CallResult and transfers. Transfers to a human (1,590), calls the same classifier marks Excused (722), out-of-area/not-offered callers (298), and intro/empty hang-ups (333) take the internal viable booking rate from 23.8% to ~30.8% with no change to bookings. Worth doing for internal truth even though marketing now leads with success.
  5. Run the attribution-window analysis. We count only jobs created during the call flow; a caller the office phones back an hour later reads as a failure. Joining unbooked viable callers against CRM jobs created within ~7 days would show how many bookings we cause but don’t credit — likely the largest single gap versus competitors’ booked-rate claims.
  6. Keep working the real conversion gap. 8,392 genuinely viable calls didn’t book: no suitable slot (1,766), price objections (898), wanted a human (450). These move revenue, not just metrics.