Which costs you more: the call you missed, or the one you answered badly?

Open your phone's call log and the misses are right there, easy to count and easy to feel bad about. The call your office picked up on the fourth ring, took a first name from, and promised a callback on sits in the same log as a success. One of those leaks you can see. The other one you mostly can't.

Here's the short answer. On its own, a missed call nobody returns costs more than a badly answered one. It can't book at all, and a fumbled call still books some of the time. But a bad answer can cost more than silence when it books the wrong job, takes down the wrong address, or promises a time nobody can keep.

Which leak costs you more in total depends on your own counts, and no published data compares the two call for call. What the benchmarks do show is that both are big. In Invoca's 2026 home services report, 65% of calls longer than 15 seconds reached a person, and the rep asked for the sale or appointment on 45% of the calls Invoca scored.

Full disclosure: Willison sells an AI receptionist that answers calls and handles them, so it sells a product aimed at both leaks and neither answer is neutral for us. Invoca sells call analytics and AI voice agents, so it isn't neutral either. That's why the method below runs on your own call log and your own calls, and why two of the fixes in this post cost nothing.

What happens to the same caller each way?

It's the first hot Thursday of the year. A homeowner's AC quit overnight, the upstairs is miserable, and at 10:40 they find you and call.

Version one: nobody picks up. Your office person is on the other line and you're on a roof. It rolls to voicemail. Invoca's platform data found fewer than 3% of callers sent to voicemail leave a message, so what you'll likely see later is a number with nothing attached.

Version two: somebody does pick up. It's the desk on a morning with two lines ringing, or it's you, answering from the ladder. You get a first name and "the AC's out," say someone will call back to set a time, and hang up to grab the other call. No address, no system type, no time offered. The callback goes out after lunch.

Both versions can end with the job going to whoever offered a time first. The difference is what your records say afterwards.

Why does only one of them show up in your call log?

A call log records whether a call was answered and how long it ran. Version one is a miss, with a number you can ring back. Version two is "answered, three minutes", exactly like a call that booked. A booking system gets you a step further: it shows which answered calls didn't book, but not why.

Plenty of unbooked calls are nobody's fault. The caller was out of your area, shopping price, or wanted work you don't do. Telling those apart from a fumble takes the call itself, a recording or someone listening. An owner who watches only the misses is measuring the leak that's easiest to see, not necessarily the bigger one.

How big is each leak in the benchmarks?

Invoca's Home Services Lead Conversion Benchmarks Report 2026 measures both. On the missing side, the report found a person answered 52% of all calls to home services businesses. Leave out the ones under 15 seconds, mostly misdials and hangups, and that rises to 65%. Past 30 seconds, it's 73%.

On the handling side, Invoca scored answered calls, which it describes as phone leads, against a set of checks. On the calls it scored, the rep:

• gave a proper greeting on 81%
• gave a proper closing on 72%
• got the caller's information on 62%
• asked for the sale or appointment on 45%
• converted the lead to a booked job on 45%, the same figure for a different check

And 58% scored as an excellent call overall. The flip side of that ask rate: on the majority of calls Invoca scored, 55% of them, the rep never asked the caller to book.

Three cautions before you lean on any of it. These are averages across Invoca's own customer base, businesses that track their calls with Invoca, so they describe other companies' phones, not yours. A checklist score also isn't a booking: a rep can skip the formal ask with a caller pleading for the first open slot, and still book it.

And the two leaks are measured on different sets of calls. The answer rate covers every inbound call, misdials and junk included. The handling scores cover answered calls Invoca counted as leads. You can't add them into one number, and you shouldn't set the miss rate against the handling scores as if they were the same kind of figure.

When does a bad answer cost more than a missed call?

The per-call ordering at the top is logic, not measurement. It holds when the bad answer simply fails to book. Some bad answers do more damage than that.

• It books the wrong job. A truck rolls for work you don't do, or for a repair that needed a different tech, and a slot on the board is gone.
• It takes down the wrong address or number. The visit's on the calendar and nobody can find the house, or the confirmation goes to a dead line.
• It promises a time nobody can keep. A caller told "someone'll be out this afternoon" who then waits all day has gone from a lead to a complaint, and maybe a public one.

Capacity moves both sides, too. In the peak week, with the board full three days out, a missed call is worth less than it is in a quiet week, though not nothing: that caller may be a replacement or a maintenance customer who'll go to whoever offers Friday. Both leaks are worth what your schedule can actually take on.

How do you measure both on your own phone?

Take one ordinary week. Not the first heat wave, and not the dead week after the holidays. The goal is two numbers covering the same week: jobs the misses cost, and jobs the handling cost.

For the misses, pull every unanswered inbound call from the log. Take out the ones you can explain without guessing: numbers that rang back and got through, known spam, vendors and the supply house, and misses somebody returned and booked. Leave an existing customer in unless you know why they called, because a regular with a new breakdown is about the best lead there is.

What's left is calls nobody booked, and you can't tell a price shopper from a real job by looking at a number. Misses you returned that didn't book, sort by why, the same way you'd sort an answered call; a callback that got fumbled counts on the handling side.

The unreturned ones, ring back and ask what they needed. If that's too many, estimate: multiply them by the share of real job requests among the answered calls you listen to below. Don't filter misses by length the way Invoca filters calls overall, because a call that rang out has ring time, not talk time.

For the handling, listen to 10 to 20 answered calls from the same week. Use recordings if your phone system keeps them, or sit beside the desk for a morning. Rules on recording calls and telling callers vary by state, so check yours before you switch recording on. Score each call against four plain checks, which cover the same ground as Invoca's scorecard:

1. Greeted with the business name. The caller knows they reached the right shop.
2. Got the name, callback number and address. Enough to call back and find the house.
3. Asked for the appointment, or offered a time. Not "someone will call you."
4. Ended on a next step the caller could repeat. What happens, and roughly when.

Count the real job requests among them that didn't book because of how they were handled. Then scale that up to the whole week: fumbles divided by the calls you listened to, times every answered call that week. Without that step you'd be setting a full week of misses beside a sample of 15, and the misses would win every time.

Last, turn both counts into jobs the same way. Even a well-handled request doesn't always book, so multiply each pile by the share of well-handled job requests in your sample that did book. It's the same rate on both sides, so a rough one won't change which pile is bigger. Now both numbers are jobs lost in the same week, measured the same way. If you want the miss side as a yearly figure, how many jobs a year unanswered calls cost walks through it.

What does it look like for a restoration company?

In Invoca's 2026 benchmarks, restoration lands seventh of the nine home services groups on answering, with a person picking up 51% of calls. Of the calls that got answered, 34% were leads, and 30% of those leads booked before the caller hung up. A job that books on the callback isn't in that 30%.

On the calls that ring your own line directly, the homeowner or property manager with water on the floor, handling has traps of its own.

The details are the job. A call that ends without the source of the water, whether it's still running, which rooms are wet, the address, whether the caller owns the place or rents it, and whether a claim's already open hasn't given the team enough to send the right crew. That's a fumble even if the caller hung up happy.

The handoff is part of the answer. A water call taken well that then sits in a notes field for an hour has been fumbled after the fact. EPA's mold guidance says to dry water-damaged areas and items within 24 to 48 hours to prevent mold growth. A slow handoff costs more here than it would on a tune-up.

Then there's the insurance question. Callers ask whether they're covered. Whoever answers can't promise coverage either way. The good answer is that the insurer decides, and that the details they're giving now will help whoever looks at it.

Which should you fix first?

Set your two weekly numbers side by side. If the miss number is bigger, start there. If the handling number is bigger, the room is in how calls get handled, and the cheapest fix there is free.

Here are the options, each with what it costs, what it's good at, and where it breaks.

• Ask for the job, every time. Free. Fixes the handling side directly: put the four checks on a card by the phone and listen to five calls a week. Breaks when nobody keeps checking that it's being done.
• Return every miss fast. Free. A rule that missed calls get a call back within 15 minutes, and a voicemail greeting that says when they'll hear from you. Breaks on the busiest days, when the people who'd call back are the ones who missed the calls.
• Text missed callers back automatically. Cheap. Answers the miss in seconds, in writing. Business texts can need registering with the carriers to avoid being filtered, and it breaks when the caller has already reached someone else.
• Roll missed calls to a second person. Costs that person's attention. Keeps a human voice on the line. Breaks when they're tied up too.
• Cover the gaps in-house. Costs payroll, say a part-timer for lunch and late afternoons. Someone who knows your shop. Breaks on nights, weekends, and the day they're out.
• A human answering service. Costs a monthly fee. A trained person answers at any hour. Only as good as its script, and some services take a message instead of booking.
• An AI receptionist like Willison. Costs a monthly subscription, ours on the pricing page. Willison picks up in seconds at any hour, and the tenth caller is answered as fast as the first. Only as good as its setup: a wrong service list or calendar rule means wrong bookings. It can get an address wrong, some callers won't talk to it, and it won't quote a price or sell a job the way you would.

With any option that isn't you, ask how the calls that need judgment reach a person: a possible safety hazard, an upset caller, a regular who wants their own tech. Get the answer before you sign, not after the first one.

Whatever you pick, run the same week's count a month later, with the same size sample, and see which number moved. To hear how the handling side sounds on our end, press play on a real call on willisonhq.com, where our HVAC receptionist takes a call from first ring to goodbye for a sample HVAC shop.

Frequently asked questions

Not on its own. Some people call around before they choose, and that's their right. It becomes a fumble when the caller who wasn't ready leaves without a next step: no estimate visit offered, no time to follow up, and no name and number taken so anyone could. A caller who says no to a clear offer was handled fine.

As fast as you can reliably manage, and the same way every time. There's no clean published figure for how long a homeowner with a broken system waits before dialing the next shop, so any exact cutoff is a guess. Fifteen minutes is a target a small office can actually keep. Whatever you choose, write it down and check it against the log once a week.

Someone else, if you can. Whoever took a call remembers what they meant to say, not what the caller heard, and tends to grade kindly. A partner, a spouse who knows the business, or a trusted employee scoring a handful of calls against the same four checks will give you a straighter number.

Not by itself. Picking up more calls shrinks the miss pile, but a call that's answered and then fumbled still doesn't book. Whatever answers your phone, a person, a service or an AI receptionist, judge it on whether it takes the details and asks for the job, not only on whether it picks up.