How much an auto shop loses on missed calls
Noon, the lift is occupied, hands elbow-deep in grease. The shop phone rings for the second time in ten minutes. Nobody can pick up: the other mechanic is out on a job, and you don't usually keep a dedicated receptionist for three bays — it doesn't pay for itself.
The call drops. Forty minutes later you wipe your hands, see the missed call, and think you'll call back later. Then a client shows up for diagnostics, then a supplier, then the day is over.
In the evening you finally dial back. The answer: thanks, I already got it fixed elsewhere.
Why nobody counts this loss
A missed call doesn't show up in the till or in the monthly report. The report only has the people who got through and showed up. It leaves a trace in exactly one place — the carrier's call log — and almost nobody ever opens that.
Because of this, the conversation about losses stays at the level of gut feeling. "We're fine on that front," "eh, happens a couple times a day." Gut feeling almost always lies in your favor: you remember the people who showed up, and about the rest you simply know nothing.
Counting takes about ten minutes, and the number you get will be yours, not something from someone else's article.
How to pull your own number
Log into your carrier's account and pull the call log for your work number for last month. It shows every incoming call, including the ones nobody answered. Count them. Separately, check the ones that came in after closing time and on weekends.
Now some arithmetic. The numbers below are made up, just to show the logic — you'll plug in your own.
Say six missed calls a day, twenty-two working days — that's a hundred and thirty-two calls a month.
Some people call back themselves or wait for your callback. Nobody knows exactly how many; I'll take half, but that's a guess, not a measurement. That leaves sixty-six people who walked away in silence.
Not all of them would have come in anyway — you don't convert a hundred percent of inquiries into bookings. Let's take a third, that's twenty-two people. At an average ticket of 6,000 rubles that comes to 132,000 rubles a month, and around 1.6 million rubles a year.
Three assumptions in a row, and each one can be argued with. So don't trust my arithmetic — plug in your own shares and your own ticket size. If you count by margin instead of revenue, the number drops noticeably, but it's usually still substantial.
Other people's numbers, so you have something to compare against
I don't show my own clients' numbers here. These are from companies selling solutions to the same problem. They have an incentive to inflate them, and they don't show their methodology, so treat this as a reference point, not a benchmark:
- a dental clinic: out of a hundred and twenty calls, thirty-five go unanswered — that's twenty-nine percent, with losses estimated at 11.9 million rubles a year (a write-up on vc.ru, Feb 5, 2024);
- a call-tracking service describes a clinic where "496,000 rubles were quietly disappearing" — though without saying whether that's per month or per year (Silent Listener);
- an auto shop in Kazakhstan: before automation they were losing twenty-five to thirty bookings a week, because in the evening a customer writes in and gets no answer (CRM AI, Jan 1, 2026).
The numbers vary, but they point the same direction, and it usually matches what people see in their own call log.
What doesn't work
Calling back in the evening. By evening the person has already solved their problem, and you hear a polite "thanks, no longer needed."
An answering machine saying "your call is very important to us." It only confirms to the caller that they can't reach you, and they keep calling down their list.
What works
Give the person a second way to reach you besides the call, and make that way work on its own.
The missed call turns into a message: "You called the shop, I'm here, how can I help?" The person replies by text — often that's more convenient anyway, their kid is asleep or they're at work themselves. The assistant asks about the car and what happened, quotes a price range from your price list, offers an open slot, and books it. You crawl out from under a car and see: 2015 Logan, knocking noise up front, booked for tomorrow at ten.
A word on how it actually works, so this doesn't sound like a promise out of thin air. The message goes out as an SMS to the same number that called, and the conversation then moves to Telegram via a link in that SMS. I don't set up gray/unofficial WhatsApp — numbers there get banned in batches, bot included.
The bot understands voice messages too. It transcribes them and replies in text, so "hey, how much for a timing belt on a Solaris" said out loud while driving still gets through. For an auto shop that's not a minor detail — half the messages there come in as voice.
And a word on the limits. Some people only ever call and never text, usually older customers. This won't save those calls — for them you need a live person on the phone or a voice bot, and that's a different price and a different conversation.
What it costs, and what you'll need to do
From $1,000, paid once, the code stays yours, twenty days of guarantee against an agreed spec.
Now for the part these articles usually leave out.
Your time is needed too: putting your price list into a spreadsheet once, and about an hour and a half answering my questions. After that the price list lives in that same spreadsheet, and you change prices yourself — no need to come back to me for that.
And the bot has running costs on the side that I don't pocket. I worked them out so I wouldn't have to hide behind "it depends on load."
One eight-message conversation costs around 5 rubles — that's the model's official API pricing. Three hundred conversations a month is about 1,500 rubles. Voice transcription adds another 250 rubles or so. The server costs about $6 a month, and it's not limited to just one bot.
All told, under normal load, that's roughly 2,000 rubles a month, and you pay it directly to the providers, not to me. If your volume is half that, the cost is half too — it scales with the number of conversations, not a figure I made up.
Pull last month's call log and do the math. If the number comes out smaller than the cost of the work, I'll be the one to tell you it's not worth doing.
Get your own number — loss calculator, three figures from your own call log.
Website and business audit — your site or app address, business name and city, I'll send a written report with a plan.