What counts as a bookable call?

The denominator is the biggest fairness problem in call QA

Two reps with identical skill can show very different booking rates purely because one sits on the queue that absorbs the vendor calls and the wrong numbers. Nobody argues about the numerator; the arguments are always about what was counted as an opportunity.

This is also the easiest thing to fix, and the most commonly skipped. Write the exclusions down, apply them mechanically, and let reps see how each of their calls was classified.

There is a second-order effect worth naming. When reps believe the denominator is wrong, they stop trusting every number built on top of it, including the ones that are fine. Denominator hygiene is therefore a prerequisite for coaching, not a reporting detail.

The standard exclusions

• Not a customer. Vendors, suppliers, job applicants, sales calls, wrong numbers, silent and dropped robocalls.
• Already handled. Confirmations, reschedules, arrival-time questions, callbacks about an open job.
• Out of scope. A service you genuinely do not offer, or an address outside your service area.
• No opportunity. Calls under a few seconds, hang-ups before a person answered, and pure billing questions.

The gray cases are where policy has to be explicit

Some calls sit on the line. A caller just outside the service radius who could be booked with a trip charge. A homeowner asking about a service you technically offer but rarely run. An existing customer calling about a warranty issue that may or may not become a paid job.

There is no universally correct answer, but there is a wrong process: deciding case by case after the fact. Pick a rule, document it, and revisit it quarterly. A written rule you disagree with produces fairer comparisons than an unwritten rule everyone applies differently.

One practical rule helps: whoever writes the exclusion policy should have to apply it to fifty real calls before publishing it. Most ambiguity surfaces in the first twenty and gets resolved in an afternoon.

Classification quality decides score quality

Once every call is transcribed, classification stops being a manual chore and becomes something the system does on every call: intent, service type, whether the caller was in area, whether an appointment already existed. That is the unglamorous half of automated evaluation and it matters more than the scoring model.

It is also where an integration earns its keep. Knowing whether the caller already had an appointment requires reading the schedule, not the audio — which is why call analysis that is wired into your operational system classifies more accurately than a standalone recording tool ever can.