ScheduledHVAC
    Operations

    HVAC Lost-Lead Callback Audit: Ask Why They Did Not Book

    Ren Álvarez circles 57 unbooked inquiries. "Fifty-seven people didn't book. Those were bad leads." It is a reasonable suspicion when the acquisition bill arrived and the appointments did not. An unbooked inquiry does not tell you why it was lost.

    Ren is an illustrative owner, not a client case study. In her recurring June story, 22 of 33 install inquiries did not receive a callback within an hour. Her first reaction had been to change targeting. The timestamps made that answer less comfortable.

    A callback audit asks the people behind the missing appointments what happened. It can expose a broken promise or a misunderstood request. It cannot turn a handful of answers into certainty about everyone who stayed silent.

    Ren needs a short research assignment, a few neutral questions, and a place to put the answers before she buys a different explanation.

    Which unbooked inquiries should you investigate?

    Start with a defined intake group and cutoff. For example, use a separate hypothetical group of eighty distinct new-job inquiries received during one stated period. Twenty-three booked by the review date, leaving fifty-seven unbooked. These are modeled audit counts, separate from Ren's June history.

    Allow enough time for the decision being studied. A replacement prospect still comparing proposals is not automatically a lost sale. Separate open decisions from confirmed losses, and record the cutoff you used.

    Clean up the list before choosing whom to contact. Confirm repeat contacts are attached to their original inquiry, identify requests already resolved under another job number, and separate work outside your territory or scope. Keep those exclusions visible in the original intake reconciliation.

    Choose a spread of cases rather than only the easiest names to reach. Include different intake times, sources, initial service or install intent, and recorded outcomes. If the sample is deliberately targeted at one failure, say so. It is a diagnostic sample, not a random portrait of the entire business.

    The call tracking and attribution setup helps connect the original request to later calls and jobs. Check those records first so you do not ask someone why they never hired you when your team completed the work yesterday.

    Ren's fifty-seven becomes a research queue with boundaries. She can see which questions the records already answer and which need a homeowner's perspective.

    How do you ask without turning research into another sales call?

    Choose a calm interviewer who can listen without defending the office or negotiating a sale. A person who did not handle the original inquiry may find it easier to hear an uncomfortable answer.

    Use a suitable contact channel for the original request and the person's stated preferences. Identify the business, explain the purpose, ask permission, and make declining easy. Stop when someone declines or asks for no further contact. This audit does not require recording calls.

    A simple opening is enough:

    "Hi, this is Alex with Álvarez Air. You contacted us about your system recently. I'm checking how we handled the request, not calling to restart a sale. Would you be willing to answer two quick questions? It's fine if now isn't a good time."

    Then ask open questions before offering categories:

    • "What were you hoping we could help you with?"
    • "What happened after you first contacted us?"
    • "What mattered most when you decided what to do next?"
    • "Was there anything we made difficult or left unclear?"
    • "Is the project resolved, still under consideration, or no longer needed?"

    The SBA's market research guidance includes direct customer interviews as a way to understand a specific audience. That supports asking; it does not make a small callback sample statistically representative.

    Ren's interviewer has one assignment: learn what happened. If a homeowner independently asks for help, handle that as a separate service request with an appropriate handoff. Do not make another sales attempt the price of answering the research questions.

    What should the interviewer record?

    Keep the record short enough to finish accurately. You need the inquiry ID, intake date, initial intent, source where known, original handling, interview attempt, whether the person answered, whether they agreed to participate, and what they said.

    Capture the customer's own wording before choosing a code. Then assign one primary reason for this audit and, where useful, a secondary factor. Define the code so two reviewers would make broadly similar decisions.

    Useful primary categories include unavailable appointment, response delay, scope mismatch, price or proposal concern, unresolved timing, chose another provider for another stated reason, and unknown. Do not use "price" merely because someone selected a competitor.

    Separate three kinds of evidence:

    • Observed record: the first callback occurred the following afternoon.
    • Customer report: the homeowner says another company offered an earlier visit.
    • Reviewer hypothesis: the delayed response may have affected the decision.

    Those statements can support an investigation without proving the same thing. A homeowner may remember the sequence imperfectly. Your own notes may also be incomplete.

    Use the call handler training process when the record reveals an unclear explanation or no appointment offer. A research code becomes useful when it points to a specific behavior you can inspect.

    Ren can now read a line that says "earlier appointment elsewhere" instead of a line that says "bad lead." The customer supplied the first statement. The second had been her assumption.

    What can six answers from ten attempts tell you?

    Continue the hypothetical audit. The office selects ten of the fifty-seven unbooked inquiries and makes ten initial research attempts. Six people answer and agree to participate; four do not answer. No further interview outcomes are assumed.

    Of the six participants, suppose the primary reasons are:

    Primary stated reasonInterviews
    Available appointment was too late2
    Response arrived after they selected someone else2
    Requested work did not fit the offered service1
    Project timing changed1
    Total participating interviews6

    The totals reconcile: ten attempts, six interviews, four unanswered. The six primary codes sum to six. Four participants mentioned either appointment availability or response timing, but those remain two distinct problems.

    You cannot conclude that four-sixths of all fifty-seven inquiries were lost for those reasons. People who answer may differ from people who do not. Ten selected cases may also differ from the rest of the queue.

    The answers justify checking appointment promises and response timestamps across more records. They do not justify a companywide recovered-revenue estimate. An unanswered research call remains unknown, not a dissatisfied customer and not an available sale.

    Ren had fifty-seven explanations in her head and no interviews. She now has six reported experiences, four unanswered attempts, and a narrower investigation. That is progress without pretending it is a census.

    Which fix follows each kind of answer?

    Match the response to the actual failure. If suitable customers need an earlier appointment, inspect available capacity, promises, and routing. More inquiries cannot manufacture a safe install opening.

    If contact arrives too late, examine notifications, named ownership, coverage gaps, and the next action when the first attempt fails. Compare the operational record with what the interviewer heard before changing a process.

    If the work does not fit, inspect source targeting and the wording of the page or offer. This is especially useful for shared lead vendors, where a contractually valid request may still be a poor operational match.

    If a replacement decision is still open, use a permission-based install proposal follow-up process. Keep "considering" distinct from "lost," and stop at the customer's stated boundary.

    If several issues appear together, choose one test you can observe. Give it an owner, start date, check date, and success measure. For example, check whether eligible requests in the covered window actually receive the assigned first response. Do not claim more bookings before a comparable cohort has had time to decide.

    Ren's first action can be a response-coverage repair rather than a new campaign. The audit has not proved targeting is perfect. It has identified a known defect that a different target audience would still encounter.

    What should Ren do before buying another lead?

    Take the findings into the same operating review as the acquisition report. Use the agency audit questions to ask which source or message could contribute to the observed mismatch, while keeping office response and appointment availability assigned internally.

    Preserve the unresolved cases. A useful report can say "six interviews suggest this problem; the remaining cases do not establish why they did not book." That is more actionable than a confident label nobody can verify.

    Repeat the audit after a meaningful process change using comparable selection rules and a suitable cutoff. Treat the next sample as more evidence, not a guaranteed demonstration that the change worked.

    In Ren's illustrative review, 57 is still written on the page. Beside it sit six customer accounts and the response timestamps her team will inspect next.

    She has stopped asking the number to tell a story it cannot tell. Before the next purchase, she can name the question, the evidence, and the person responsible for changing what happens after the inquiry arrives.