Is it cheaper to buy more leads or to close the ones you already get?

It's the end of a slow month and the ad rep is on the phone with a bigger package. More clicks, more calls, a better spot on the page. Meanwhile your phone log shows a column of calls nobody picked up. You've got one budget. Do you spend it on more calls, or on booking more of the ones already ringing?

Here's the short answer. Price one more booked job both ways, with your own numbers. Buying costs your cost per call divided by the share of calls that end up booked. Closing costs whatever the fix costs, divided by the extra jobs it books out of calls you've already paid for.

When most of your calls already reach a person and most leads already book, as with plumbing companies answering 74% of calls in Invoca's 2026 benchmark, buying is often the cheaper move. When only about a third reach a person, as with HVAC companies at 34%, closing often is, as long as the fix costs less than the extra jobs are worth.

Full disclosure: Willison sells the closing side. It answers calls and books jobs, and it doesn't sell leads, so we have an interest in how this comes out. That's why the method is laid out step by step, why there's a section on when buying wins, and why our side's failure modes get their own section too. Check the math against your own phone log, not against our word.

What does each option do to the same month?

Picture the same month two ways. In the first, you buy more. More calls come in, and every one of them runs through the same phone you've got now: the same share answered, the same share that turns out to be a real job, the same share that books. Whatever leaks today leaks on the new calls too, at the same rate.

In the second, ad spend stays flat and you fix how the calls get handled. Answer more of them, ask for the job more often, call back faster. Nothing new comes in. More of what already rings turns into work on the board.

There's a third option the question leaves out, and it can be the cheapest of all: buy better. Cut the keywords that bring junk, dispute the bad leads a pay-per-call platform may credit back, and move money from a source that rings but rarely books to one that books. Willison doesn't sell that either. It just belongs on the list before you spend another dollar on either side.

How do you price one more booked job on each side?

Cost per lead is the wrong yardstick here, because it stops before the leaks. The fair comparison is cost per extra booked job. You need four numbers, and the best place for all four is your own phone log and job board. Invoca's benchmark rates are a fallback for a shop that has none yet.

• Cost per call: what a source cost you last month, divided by the calls it produced. Work it out per source, not blended, because sources differ a lot on both cost and how often they book.
• Answer rate: the share of those calls a person actually picked up.
• Lead rate: the share of answered calls that were real prospective jobs, not vendors, wrong numbers or existing customers checking on a visit.
• Book rate: the share of those leads that booked.

Buying: cost per call ÷ (answer rate × lead rate × book rate) = cost of one more booked job.

Closing: monthly cost of the fix ÷ extra jobs it books = cost of one more booked job. For a fix that answers more calls, extra jobs = newly answered calls × lead rate × book rate × (1 − the haircut). The haircut is the share of those calls you'd have won back anyway, plus the share that were never jobs at all.

That last term matters, and it's easy to skip. Not every unanswered call is a lost job. Some callers ring back. Some get a callback from the truck ten minutes later and still book. Some were misdials, robocalls or hangups that were never work. Your phone log can tell you roughly how big each of those is, so use it rather than assuming zero.

What does the math say for an HVAC shop?

Take a placeholder shop: one ad source producing 200 calls a month at $20 a call. Those inputs are invented for the example, so swap in yours. If your cost per call is higher, buying costs more per job than shown here.

For the rates, use what Invoca's Home Services Lead Conversion Benchmarks Report 2026 found for HVAC: 34% of calls answered by a person, 45% of answered calls were leads, and 43% of those leads converted on the call. Invoca counts only leads that converted on the call, so treat that last rate as a floor for your book rate.

One caution on that 34%. It counts every inbound call to HVAC companies, not one ad source, and it includes misdials and quick hangups. The same report found that once calls under 15 seconds are filtered out, the home services answer rate rises to 65%, and to 73% for calls over 30 seconds. A chunk of any unanswered pile was never a job, which is what the haircut below is for.

Multiply those through and 200 calls produce about 13 booked jobs. At $20 a call, each booked job from buying more costs about $304.

Now the closing side. Say a fix lifts the answer rate from 34% to 52%, which is Invoca's average across home services, not a promise from anyone. That's 36 more answered calls, which at the same lead and book rates comes to about 7 more jobs. Seven jobs at $304 each is about $2,100 worth of buying. So the fix is the cheaper path if it costs less than about $2,100 a month.

Then apply the haircut. If a quarter of those calls would have been won back anyway or were never jobs, it's about 5 extra jobs and a ceiling of about $1,590. If it's half, it's about 3.5 jobs and about $1,060. If your log shows you win back most of them, closing may not beat buying at all. That range is the honest answer, not the top number.

There's a shortcut hiding in this. If a newly answered call books as often as the calls you answer today, it's worth cost per call ÷ answer rate in bought calls. At $20 and 34%, that's about $59. The lower your answer rate, the more each rescued call is worth next to a bought one.

When is buying more leads the cheaper move?

Run the same shop with Invoca's plumbing rates and the picture flips. Plumbing companies answered 74% of calls, 30% of answered calls were leads, and 55% of leads converted on the call. Two hundred calls produce about 24 booked jobs, at about $164 each from buying.

Even perfect answering only reaches the 26% of calls that didn't reach a person. At the same rates that's about 8.6 jobs, worth about $1,400 of buying, before any haircut. Real fixes don't hit 100%, so the room is smaller still.

Buying also wins in situations no benchmark shows:

• You need jobs this month. Ad spend turns on and off in a day. A closing fix can take weeks to show up in the numbers.
• The crews are idle and the phone is quiet. If calls are what's short, there's nothing to close.
• A new service or a new town. You can't book calls that don't exist yet.
• You already answer nearly everything and ask for the job every time. Your closing room is close to used up.
• Both sides come in well under your profit per job. Then cheaper isn't the question. Do both, and let crew capacity set the limit.

That last one deserves a beat. These aren't always rivals. They only compete when the money is the constraint. Hold each side's cost per booked job against what a booked job actually earns you, not just against the other side.

Why does the next dollar on either side buy less than the last?

Both formulas use averages, and you're spending at the margin. Each side flatters itself in its own way.

On the buying side, your average cost per call includes the cheap calls you already get. The next tranche of spend can reach people who are less ready to call, so the next call can cost more and book less than your average one. More volume with the same staff can also push your answer rate down, because calls start landing on top of each other. That one tilts toward closing, and it's still true.

On the closing side, the first fixes are cheap and the last ones aren't. Asking for the job on every call costs some training. Covering lunch costs a schedule change. Getting from good coverage to every call costs real money. And the calls you're missing may not look like the ones you answer. After-hours callers can be more urgent. Missed calls can also carry more spam and more people who'd have rung back.

So discount both. Assume your next bought call is a little worse than average, and your next rescued call too. If one side still wins after both haircuts, the answer is solid.

What can go wrong on the closing side?

This is the side Willison sells, so here's where it can fall short.

• A ramp before it performs. Any new answering setup, ours included, has to be set up and tuned before it's live, so the gain doesn't start on day one.
• Borrowing the benchmark lift. Moving from 34% to 52% is Invoca's industry average, not your result. Measure yours.
• Callers who'd have come back anyway. Skip the haircut above and you'll overrate any closing fix, ours included.
• A monthly cost in a month with little to close. A paid fix keeps billing when the phone goes quiet. Bought calls stop costing the day you stop buying.
• Callers who won't talk to it. Some callers would rather not deal with an automated voice, and an AI receptionist can't win those back.
• Hidden costs. Owner time, callback labour, and jobs booked wrong all belong in the fix's monthly cost, the same way lead subscriptions and the time to handle extra calls belong on the buying side.

The closing options themselves run from free to paid, and each has a weak spot. Training the team to ask for the job costs nothing but practice. Covering lunch and evenings in-house costs payroll. Forwarding to the owner's cell costs nothing until the owner is on a ladder or under a sink. A human answering service puts a trained person on the line, if its script books instead of taking messages.

An AI receptionist like Willison answers every call at any hour, several at once, and books qualified jobs to your calendar. Its weak spot is that it can mishear an address or misclassify a job. Price it the same way as everything else on this list: by the extra jobs it books, after the haircut.

One more case sits outside the phone math entirely. If a lot of what you buy is form-fill leads, which can be sold to several companies at once, there's no answer rate to fix. There, the closing lever is how fast you call back. HBR's study of 1.25 million sales leads found firms that tried to reach the lead within an hour were nearly seven times as likely to qualify it as firms that tried even an hour later. It's an older study of sales leads, not home services calls, so treat it as directional.

How do you decide this month?

Pull 30 days from your phone log and your job board. For each source, count calls, answered calls, real leads, booked jobs, and jobs that actually ran or sold, along with the average ticket. Booked isn't banked: cancellations, no-shows and diagnostic visits that never become a sale all thin the number, and they don't thin it evenly by source.

Work out both costs per booked job, apply both haircuts, and hold each against your profit per job. Spend where it's cheaper. Then run it again in a different season. An HVAC shop's phone in a heat wave and the same phone in October are two different businesses, and the answer can flip between them. Crew capacity caps both sides: an extra booked job is only worth something if somebody can run it.

For more on the inputs, the calls-per-job math breaks down the three rates, cost per booked job covers what belongs in the numerator, and tracing booked work to its source is how you get per-source numbers in the first place.

Frequently asked questions

Your phone system or carrier's call log marks missed calls, and that's enough. Take one ordinary week, count every inbound call and how many a person picked up, and divide. Leave out your own outbound calls. A week won't be perfect, but it's your number, which beats any benchmark for this decision.

Nobody publishes a clean figure for how much it moves bookings, so we won't invent one. What Invoca's 2026 call-handling chart does show is that the rep asked for the sale or appointment on 45% of the home services calls it scored. If your calls look like that, asking every time is about as cheap as a closing fix gets. Measure your book rate before and after.

Not automatically. Cutting spend cuts calls, and the closing math assumes the calls keep coming. Cut the worst source first, the one with the highest cost per booked job, and only if the closing fix beats it after the haircut. If both sides cost well under your profit per job, you may not need to cut anything.

Invoca measured restoration too: 51% of calls answered by a person, 34% of answered calls were leads, and 30% of those leads converted on the call. Plug those in as a starting point. The same formulas work, but a water job's ticket and timing are different from a tune-up's, so your own numbers matter even more.