Answering service conversion rates: compare actual outcomes

01 / Buying with your own evidence

Define the event before calculating a rate.

For a home service office, use a sequence that can be checked in business records: service request received, request accepted as eligible, appointment confirmed, and job completed. A message saying a customer wants Tuesday is a request; a dated appointment accepted in your scheduling record is a different event. An appointment does not prove that work was completed.

Choose the question first. Confirmed appointments divided by eligible new inquiries measures booking within a screened group. Confirmed appointments divided by all new inquiries also includes demand you could not serve. Both can be useful, but their labels and denominators must remain visible. Neither is a universal provider conversion rate.

Google Ads documents call conversions based on a configured minimum call length. A call passing that threshold is a duration-qualified advertising event; it does not establish that an appointment or job was booked. Keep that event separate from a confirmed appointment linked to your scheduling record.

02 / Buying with your own evidence

Build one record per service inquiry, not per phone call.

A homeowner may call twice about the same broken furnace, leave a message and later confirm a visit. Keep those contacts linked to one inquiry. Count a separate new job request separately when the records support that distinction. Do not assume one telephone number always equals one person, household or job.

Create a small audit sheet with an internal inquiry ID, first-contact date, answering provider, source, requested service, eligibility decision and reason, confirmation date, job status and evidence reference. Use the information your office is authorized to review. Mark unresolved matches or missing outcomes as unknown; do not silently turn them into a loss or a booking.

03 / Buying with your own evidence

Give each group the same time to reach the outcome.

Choose a first-contact period and an observation window before calculating results. For example, a business could inspect inquiries received during a defined four-week period and allow each inquiry fourteen days to reach a confirmed appointment. That is an illustrative study design, not a recommended universal window. Choose a window that fits your own booking cycle.

Compare like groups: requested work, service area, source, staffed versus after-hours coverage and appointment availability. A booked calendar or different service mix can change results independently of who answered. If groups differ, show separate rows rather than presenting a causal claim about the answering provider.

Keep marketing credit separate from operational outcome. CallRail documents several attribution models that assign credit to different touchpoints. Switching that model can change reported lead totals without creating another confirmed appointment. Preserve the attribution setting when comparing exported reports.

04 / Buying with your own evidence

Use the comparison to locate the next operational question.

If eligible requests do not become appointments, inspect the documented reason: no suitable slot, customer declined, follow-up pending or outcome unknown. Keep a requested callback separate from a completed callback and a confirmed booking. This makes the comparison useful to the office without awarding a provider credit for work that happened later elsewhere.

Produce a dated table with the inquiry period, observation cutoff, definitions, exclusions, counts and limitations. A before-and-after comparison is observational: seasonality, marketing, staffing and availability may also have changed. Use it to decide what to investigate or test next, not to claim guaranteed improvement or estimate unobserved revenue.