What your booking rate denominator excludes

A ratio inherits the honesty of its denominator

Booking rate is a ratio: booked jobs over calls. The numerator is unambiguous — a job either exists in the field service management platform or it does not. The denominator is where the metric is actually decided, and the population it uses is a configuration choice rather than a property of the business. Two operations with identical phone traffic and identical crews can publish different booking rates purely by defining calls differently. The report that displays the ratio does not restate which population produced it. The population is a decision made once at configuration time and then inherited, unstated, by every report built on top of it.

What a field service platform can see

A field service management platform records what is written into it. A call becomes visible to it when a CSR opens a customer record, a lead or a job while on the phone, or when a call tracking integration writes an inbound event against a contact. Calls that never reached a CSR produce no such write. The platform is not concealing them — it was never told about them. That means the denominator assembled inside the platform is a connected-call denominator, and the classes of inbound contact that can sit outside it each live in a different system:

• Abandoned in queue — the caller hung up while on hold. The trace is a call detail record in the phone system or call tracking platform, carrying a hold duration and no answer event.
• Rang out — the line was staffed on paper but nobody picked up. The trace is the same call log, carrying a ring duration and a disposition code.
• After hours — the call arrived outside dispatch hours and went to voicemail. The trace is a voicemail record plus whatever transcription the system emitted.
• Answering service — a third party took the call and produced a message. The trace is the answering service's own message log, which lives outside the platform entirely and is not queryable from it.
• Overflow or busy — the call hit a forwarding path, a second office number, or a mobile that was never instrumented. Depending on the routing, the trace may not exist at all.

The direction of the error

If the phone queue degrades — a CSR leaves, a shift goes uncovered, seasonal volume spikes — the additional calls that fail land outside the denominator, not inside it. The connected calls that remain are the ones a CSR actually worked, and those book at whatever rate CSRs book at. So the reported ratio can hold steady, or rise, while the business captures less of the demand arriving at its phone number. A metric that moves the wrong way under stress is worse than no metric at all, because it is load-bearing: staffing decisions, media budgets and channel comparisons are all argued against it.

The reconciliation that settles it

1. Export the full inbound call detail record for a fixed window from the phone system or call tracking platform, including calls with no answer event, with caller number, timestamp, ring or hold duration, and disposition.
2. Export the answering service's message log for the same window, where one sits in the path.
3. Export customer, lead and job creation records from the field service management platform for the same window, including created-at timestamps and the phone number stored on each record.
4. Deduplicate the call log by caller number within a bounded window, so repeat dials from one household collapse into a single inquiry rather than inflating the denominator.
5. Join the call log to the platform records on phone number and time proximity, and classify every inbound inquiry as connected-and-booked, connected-not-booked, or never-connected.
6. Recompute the ratio across all three classes, and publish it beside the platform's stock booking report rather than instead of it — the difference between the two figures is the finding, not either figure alone.

Why the paid side cares about the same population

The denominator is not only a reporting question. Google's Local Services Ads About ad rankings documentation lists a business's responsiveness to customer inquiries and requests among the factors that feed ad rank, and beneath that bullet states: "Missed calls may negatively affect your responsiveness." The same page names average response time among the profile-quality factors ^[SRC-10]. So the population your booking rate excludes is the same population a paid placement is reading as a signal. Treating unanswered calls as not counted on the operations report while the ad platform counts them is the inconsistency worth resolving first. The rest of that path is set out in the Home Services industry record.

What the rebuilt denominator still cannot show

• An unmatched call is not a lost job. Suppliers, wrong numbers, recruiters and existing customers calling about an open ticket all appear as rows in the same log.
• A caller who never connected may have redialled later from a different number and booked. A number-level join cannot see that, and will count the same household twice.
• The join says nothing about what the job would have been worth. A never-connected call and a replacement estimate are indistinguishable in a call detail record.
• Platform-reported conversions are not a clean cross-check. Google's consent mode documentation states that when users deny consent for storage, tags send measurements without cookies, and that "Google products use these pings to model your metrics" ^[SRC-18] — so part of what an ad platform reports is modelled rather than observed.
• Nothing in the join establishes a cause. Coverage, routing, IVR depth and hold-time tolerance are separate variables that a single log cannot separate.

If the remediation is an automated text-back

One structural response to a widened denominator is to fire an automated message to every number that called and did not connect. That crosses into regulated territory and should be designed accordingly. Under the FCC's TCPA delivery restrictions at 47 CFR 64.1200, autodialed or prerecorded telemarketing calls and texts require prior express written consent; a consumer may revoke by using any reasonable method to clearly express a desire not to receive further calls or text messages; and revocation must be honored within a reasonable time not to exceed ten business days ^[SRC-14]. Whether a specific text-back is telemarketing or a response to an inquiry is a legal determination, not a systems one. The systems consequence is concrete: consent state, revocation timestamps and suppression must be first-class fields on the contact record before any automated recovery path is switched on, not retrofitted after it.

The denominator question is a measurement-layer problem, so it belongs with the rest of the tracking, attribution and measurement systems rather than with call scripting. The join described above runs on internal exports — call detail records and platform tables — which is Internal Revenue Audit work rather than scan work. What a Revenue Scan covers is the public half: the numbers a prospective customer can dial, the profile that answers for them, and the booking surfaces on the site.