ScheduledHVAC
    Strategy

    HVAC Shared Lead Vendors: What Did You Actually Buy?

    Ren Álvarez sees twenty charges on a lead invoice. "Twenty leads means twenty chances to book." That feels reasonable when every notification has a price attached. Twenty billed contacts are not necessarily twenty distinct opportunities.

    Ren is an illustrative owner, not a client case study. Her recurring story includes $34,000 spent over four months on a subscription that sold the same leads to two competitors. She kept paying after recognizing the problem because inquiries really were arriving.

    That history does not prove every marketplace is a poor purchase. It explains why she needs to identify the unit before judging its price. A ringing phone, a shared request, and an appointment already agreed are different things to buy.

    The next test should produce an answer she can use, including a reason to stop.

    What exactly is the vendor selling?

    Ask for a plain description of the billable event. Is it a notification, connected conversation, homeowner request, qualified opportunity, or confirmed appointment? What evidence establishes that event?

    Clarify sharing in the same conversation. Does the request go to several contractors? How many, where, and on what timing? What does any promise of exclusivity actually cover? Obtain the answer for the product you are buying rather than assuming a familiar word means the same thing everywhere.

    Google's Local Services guidance, for example, describes payment for valid leads. That is not payment only for completed jobs. It is one platform's defined unit, not a description of every shared-lead vendor or its credit rules.

    Request the matching criteria, territory, job types, notification route, duplicate policy, credit process, volume limit, and the procedure for changing or stopping the purchase. Keep a copy of the current terms you actually accepted. Do not build the test around a salesperson's informal promise that every lead will be good.

    Ren's old subscription supplied activity. The missing question was whether that activity matched the acquisition unit she thought she was purchasing. She can answer that question before another invoice arrives.

    What has to be ready before the first notification?

    Define the work and dates your team can serve. A source built around urgent repairs is a poor fit for an office that cannot respond during its delivery hours. A replacement inquiry also needs a different conversation and follow-up window.

    Assign an intake owner and backup. Agree how a new request is acknowledged, contacted, qualified, and offered an actual appointment. Use the call handling checklist instead of assuming the notification itself completes the handoff.

    Label the source consistently in the operational record. Attach later calls about the same request to that inquiry. Compare phone, address, requested work, timing, and the office's confirmation before treating repeat contact as a new opportunity.

    Then distinguish operational eligibility from billing validity. Your company may decide a request is unsuitable while the vendor's stated billable criteria still apply. Record the facts needed for a credit request without assuming every unbooked job deserves one.

    Keep the homeowner's experience in view. Identify your business and the request they made. If they have already selected another provider or decline further contact, update the outcome rather than having several employees continue the same pursuit.

    Ren can now tell the office what this source requires. A vendor cannot supply an extra CSR, install opening, or response system simply by supplying another request.

    What happened to twenty billed contacts?

    Use a separate hypothetical service-inquiry trial, not Ren's subscription history. Receive contacts over ten days and follow the same group through day forty. The prices and results below are assumed for the exercise.

    Suppose twenty billed notifications contain four repeat contacts, leaving sixteen distinct requests. Four of those requests fall outside the trial's agreed operational fit, leaving twelve eligible opportunities.

    The office reaches nine of the twelve. Four book. By the cutoff, three jobs are completed and one booking has cancelled. Five reached opportunities did not book, and three eligible opportunities remain unreached.

    Trial measureCount
    Billed notifications20
    Distinct requests after repeat contacts removed16
    Eligible opportunities12
    Reached opportunities9
    Booked appointments4
    Completed jobs3
    Cancelled bookings1

    The stages overlap. Do not add them together. Twelve eligible opportunities reconcile as four booked, five reached but unbooked, and three unreached. The four bookings reconcile as three completed plus one cancelled.

    The vendor invoice still needs its own reconciliation. In this example, assume two charges receive approved credits under the stated policy. That does not imply all duplicates or unsuitable requests qualify for credits from every provider.

    Ren can leave the twenty charges visible. She simply stops reading them as twenty separate prospects her team could serve. The useful acquisition count is attached to work requests, not notification volume.

    What did each completed job actually cost?

    Assume $100 per billed notification. Twenty cost $2,000. Two approved $100 credits reduce the vendor cost to $1,800. Add $400 of allocated intake and tracking cost for this trial: $2,200 total.

    At that stated scope, cost per booked appointment is $2,200 ÷ 4 = $550. Cost per completed job is $2,200 ÷ 3 = about $733. Neither figure is the original $100 billed-contact price.

    These are not market benchmarks or claimed results. The point is to compare the same cost scope with the outcome you need. Keep pending credit claims separate from credits actually approved, and distinguish a credit balance from money already returned to your bank.

    The SBA's finance guidance describes comparing costs and benefits over a specified period. Include the source's real handling burden and compare completed-job contribution, not just its invoice value. Keep company fixed overhead and profit separate from job contribution.

    Use call tracking and attribution to reconcile the requests and outcomes. Compare service with service and similar response windows. An exclusive request at one price and a shared notification at another are not interchangeable products.

    Ren's next conversation can be about $733 per completed modeled job rather than whether $100 sounded cheap. The higher cost per completed job is a question to investigate, not proof of fraud or proof that every alternative will be better.

    How do you run a test that can answer a question?

    Choose a narrow assignment: one work type, defined territory, response coverage, intake period, observation cutoff, and spending ceiling. Record fit criteria and credit rules before results arrive.

    Evaluate source quality and your handling separately. Wrong territory suggests a matching issue. Eligible requests left untouched suggest a response issue. Suitable homeowners declining because of availability suggest a delivery issue. Do not put every loss under one label.

    The Google Ads expectations guide makes the same distinction between acquisition activity and scheduled work. Ask your vendor for evidence appropriate to its promised unit, and supply your own office outcomes where the vendor cannot observe them.

    For a small sample, inspect the individual records. A few completed jobs can change the average sharply. Do not present a short trial as a reliable forecast for every season or assume the source caused every eventual sale.

    When the reason remains unclear, a respectful lost-lead callback audit can supply hypotheses. Keep that research separate from repeatedly trying to sell someone who has already declined.

    Ren can test the source without promising to keep it. The review date is part of the purchase decision, not an admission that the first decision was wrong.

    When should Ren keep, change or stop the source?

    Keep it when the verified requests fit, the office can handle them, the economics make sense, and the result survives a suitable observation window. Change the setup when a specific correctable problem is supported by the records.

    Stop or limit the purchase when its current fit or economics do not justify more demand, following the terms and operational process already established. Money spent last month is not a reason to purchase another unsuitable month.

    Use the agency audit questions to keep reporting accountable even when several vendors touch the same inquiry. Nobody should need to choose which dashboard to believe without the underlying records.

    A lead vendor sells a defined entry point. Your completed-job cost tells you what that entry point became.

    In Ren's illustrative test, the invoice stays beside the dispatch outcomes. Twenty charges, sixteen requests, twelve eligible opportunities, four bookings, three completed jobs. Each number has a job.

    The subscription no longer gets renewed merely because something arrived. The next purchase has to earn its place in the operating plan.