A garage brings more customers back when each reminder is timed to the car's real due date, worked out from its recent mileage, and sent about four weeks ahead with the car's make, model and last work mentioned, plus a booking link. AI writes those personal messages in bulk and sorts the replies; a spreadsheet does the date maths.
The usual reminder, a fixed text eleven or twelve months after the last visit, is wrong for most cars. A delivery driver's van can pass its mileage limit in seven months; a retired customer's small hatchback may do 3,000 miles a year and need reminding on time, not miles. Get the date right first; better wording only helps once the timing is right.
What your garage system already knows about each car
Most garage management systems store everything you need, even if nobody has exported it. Pull a list of every vehicle seen in the last three years with these fields:
- Customer first name, preferred channel (text or email), and whether they've opted out
- Registration number, make, model and year
- Date and mileage of the last service, and of the visit before that (any visit with a mileage reading counts)
- The service interval you follow for that vehicle, in miles and months
- Advisories or declined work from the last visit, such as "rear pads 4 mm" or "front tyres near limit"
Expect the export to be messier than you'd like. The fields AI helps clean before anything is sent are mileages entered in the wrong box, a mileage lower than the previous visit, duplicate customers with slightly different names, and vehicles that have obviously been sold (a new owner's name against the same registration). Ask an assistant to flag those rows, not fix them silently, and correct them yourself.
Working out when each car is actually due
Don't ask a chat assistant to calculate dates for 1,500 cars; it will make arithmetic slips you won't spot. Put the logic in a spreadsheet, where it is checkable. With the previous visit date in A, previous mileage in B, last service date in C, last service mileage in D, interval miles in E and interval months in F:
G Miles per day =(D2-B2)/(C2-A2)
H Due by mileage =C2+E2/G2
I Due by time =EDATE(C2,F2)
J Due date =IF(A2="",I2,MIN(H2,I2))
K Send reminder on =J2-28Format H to K as dates. Column J takes whichever limit comes first, and falls back to the time limit when there's only one mileage reading.
A worked row, for illustration. A van was serviced on 10 March 2026 at 48,200 miles; its previous visit, on 12 September 2025, recorded 41,900. That's 6,300 miles in 179 days, about 35 miles a day. On a 10,000-mile interval it reaches the limit roughly 284 days after the service, around 19 December 2026, nearly three months before the twelve-month date. A fixed annual reminder would arrive in February, long after the van went past its service mileage. The sheet sends it on 21 November instead.
When the mileage prediction goes wrong
Two kinds of row break the formula, and both are common. The first is two readings too close together. A hatchback serviced on 2 June at 31,000 miles came back on 12 June with a puncture at 31,400: 400 miles in ten days, 40 miles a day. The sheet predicted the next service in about eight months, when the owner's real average, over the year before, was under 15 miles a day. Short trips distort the rate, so only use a previous visit if it was at least 60 days before the service; otherwise fall back to the time limit. The second is a reading lower than the one before, from a typo or a replaced instrument cluster. That gives a negative miles-per-day figure and a due date before the service date. A guard column catches both:
L Use mileage? =IF(OR(A2="",C2-A2<60,D2<=B2),"NO","YES")
J Due date =IF(L2="NO",I2,MIN(H2,I2))Filter column L for "NO" once a month and look at the rows where D2 is less than or equal to B2. Those are data-entry problems the desk can fix at the car's next visit.
Reminders that sound like you know the car
Once the sheet says who is due, AI turns each row into a message. The difference between a generic and a specific reminder is easiest to see side by side.
Before:
Dear Customer, your vehicle is due for a service. Please call us to book.
After, generated from the van's row and lightly edited:
Hi [first name], going by your recent mileage, your Transit Custom will reach its 10,000-mile service around 19 December. We have space the week of 1 December; it's usually a three-hour job, so you can wait or drop it off. Last time we noted the rear pads at 4 mm, so we'll measure those again. Book here: [link]. Reply STOP to opt out.
The prompt that produced it, which you can reuse row by row or on a whole pasted table:
Write a service reminder text (under 320 characters) for each row.
Use: first name, model, due date, the reason it is due (mileage or
time), and at most one advisory from last visit, stated exactly as
written. Friendly and plain; no exclamation marks; no warnings about
safety or breakdowns. End with "Book here: [link]. Reply STOP to opt
out." If any field is blank or looks wrong, write CHECK instead of
a message.That last instruction matters. In an illustrative first run on 200 rows, the assistant marked 11 as CHECK: blank models, a mileage of 4,800 on a ten-year-old car (a missing digit) and two customers with no first name. Those are exactly the rows that would have produced embarrassing messages.
Different customers need different reasons. A low-mileage car due on time might get: "Hi [first name], your Yaris hasn't done many miles this year, but it's twelve months since its service on 14 October, and most manufacturers set a time limit as well as a mileage one." A business customer with three vans is better served by one message listing all three due dates than by three separate texts. For an illustrative electrician with a small fleet, the grouped email reads:
Hi [first name], here's when your three vans are next due, based on their recent mileage: the white Transit (ending 4KD) around 18 November, the Berlingo (ending 7PX) around 9 December, and the Vivaro (ending 2LM) by time on 20 January. If it helps to keep two off the road on the same day, we can do the Transit and the Berlingo together in the week of 11 November, both back by 4pm. Reply with a day that suits or book here: [link].
Grouping needs one extra field in the export, the account or company the vehicle belongs to, and one line in the prompt: "Where several vehicles share an account, write one message listing each vehicle with its due date, sorted by date." Fleet customers tend to plan a few weeks ahead, so send theirs at six weeks rather than four.
The reminder calendar, from four weeks out to eighteen months lapsed
| When | Message | Channel |
|---|---|---|
| 28 days before due | Personal reminder with due date, reason and booking link | Customer's preferred channel |
| 7 days after, no booking | Short nudge: "still happy to fit you in, next free slots are…" | Same channel |
| On the due date, if advisories exist | One line about the specific advisory, no alarm | Text |
| 6 months overdue | "We haven't seen your car in a while" with an easy way to say it's been sold | |
| 18 months since last visit | Final check-in, then move to inactive |
Stop the sequence the moment a booking appears in the diary. If your system can't tell the spreadsheet about bookings automatically, the person on the desk marks the row "booked" as part of taking the booking; that one habit prevents most of the annoyed replies. For advisories that the customer actively declined, the timing is shorter and the wording different, which following up unapproved repair estimates covers. Customers who have drifted away for more than a year respond better to a different kind of message altogether; winning back lapsed customers with personalised emails goes into that.
Sorting the replies: bookings, sold cars and opt-outs
Once reminders go out in volume, replies arrive at all hours and in every style. An assistant can read each one, put it in a category and draft an answer for the desk to send or adjust. Illustrative replies and how they're handled:
| Reply | Category | Drafted response or action |
|---|---|---|
| "can i bring it thurs am" | Booking request | Desk checks the diary; draft confirms Thursday 8:30 drop-off |
| "how much is it this time" | Price question | Draft gives your standard service price for that model from your price list, never a guessed figure |
| "sold it in may" | Vehicle gone | Thank them; mark the car inactive; ask if they'd like reminders for their new car |
| "had it done at the dealer" | Serviced elsewhere | Thank them; ask for the mileage so the next reminder is right |
| "STOP" | Opt-out | Opt out immediately; no reply needed beyond any automatic confirmation |
Give the assistant your price list and opening hours, and tell it never to state a price or time that isn't in them. Prices for a service vary by model and by what's found, so the draft should give the standard figure and say it's before any extra work. The "sold it" replies are worth more than they look: each one removes a dead record, which makes every later reminder run cleaner.
The reply that needs most care is the one that describes a symptom. An illustrative case from a first week:
Customer: "yeah can do next week, its making a grinding noise when i
brake now tho, is that the pads you said?"
Drafted response: "Yes, that'll be the rear pads we noted at 4 mm last
time. We'll replace them at your service next week."The draft diagnosed a fault from a text message and committed the garage to a repair nobody had priced. Grinding under braking may be worn pads, or something else, and if it is metal on metal it shouldn't wait a week. The desk rang the customer instead and booked the car in the next morning for a brake check. The prompt gained a rule: "If a reply describes a noise, warning light, smell, leak or change in how the car drives, don't suggest a cause. Label it SYMPTOM, and draft only: 'Thanks for telling us. We'd like to have a look sooner; we'll call you today.'" Every SYMPTOM reply goes to a person the same day.
An illustrative four-bay garage, before and after
Picture an independent four-bay garage with about 1,500 active vehicles, meaning seen in the last 18 months. Around 125 cars fall due each month. Suppose its fixed annual texts bring back about 3 in 10 of those within six weeks, roughly 38 cars, at an average service invoice of $260.
After switching to predicted due dates, personal messages and one nudge, imagine that rises to 4 in 10, about 50 cars. The extra 12 cars a month are worth about $3,100 in service invoices, before any repair work found on the ramp. Your own figures could be higher or lower; the useful thing is that you'll be able to measure the change because every reminder now has a date and an outcome.
The costs are modest. Text messages typically cost from under a cent to several cents each, depending on the provider, the country and the message length, so 250 texts a month is a small line. Drafting 125 messages takes an assistant minutes; reviewing them takes the desk about 20 minutes a week. Setup, including cleaning the export and building the sheet, takes a day spread over a couple of weeks. If you're still deciding whether any of this is worth doing, the decision checklist for small auto repair shops helps you weigh it against other jobs.
Consent, frequency and tone
- Service reminders and marketing are different. A reminder that a customer's own car is due is usually treated differently from a promotion, but rules on marketing texts and emails vary by country. Keep promotions out of reminders, and check the rules that apply to you with an adviser if you're unsure.
- Every message offers an opt-out, and opt-outs take effect the same day.
- No fear selling. "Your brakes could fail" in an automated text reads as a scare tactic even when it's true. State the measurement, as the technician recorded it, and let the customer decide.
- Two reminders per due date, maximum, plus the lapsed messages. More than that and opt-outs climb.
Checking the reminders are earning their keep
Add an outcome column and review it monthly: booked within 30 days of the reminder, booked later, serviced elsewhere, vehicle gone, no response. Four numbers tell you most of what you need:
- Return rate within six weeks of the due date, compared with last year's.
- Lead time: how many days before the due date people book. If most book in the last week, send the first reminder later.
- Opt-out rate per 100 messages. A jump means the wording or frequency is off.
- CHECK rows per run. Falling numbers mean your front desk is recording mileage and names properly.
A filled-in month for the four-bay garage above, illustratively: 125 reminders sent. 41 booked within 30 days, 9 booked later, 6 serviced elsewhere, 8 vehicles gone, 61 no response. Median lead time was 9 days before the due date, opt-outs were 2, and CHECK rows were 7, down from 11 in the first run. Two decisions follow. With most bookings arriving in the last fortnight, the four-week reminder is early for this garage's customers, so the next month tests three weeks for private cars and keeps six weeks for fleets. And the 61 with no response get the day-7 nudge as planned, but the six serviced elsewhere each get a one-line thank-you asking for the mileage, which keeps them in the sheet for the next due date instead of lost.
Once customers are back on the ramp, keeping them informed while the car is in is the next step; repair status updates customers appreciate covers that, and for customers who book but don't show, reducing no-shows with reminders and rebooking fills the gap.
Garage reminder questions
Should reminders go by text or email?
Use whatever the customer chose when they booked, and default to text for service reminders because they are short and time-sensitive. Email suits the longer lapsed-customer message with a service history attached. Whichever you use, include a booking link or a number that is answered, and an easy way to stop the messages.
What if we don't record mileage at every visit?
Start recording it now; it takes seconds at check-in and every future reminder depends on it. Until you have two readings for a car, fall back to the time-based due date from the last service. A year from now most of your regular cars will have two readings and the mileage prediction will cover them.
Can the AI book the car into the diary itself?
It can if your garage system or booking tool has an online diary it can write to, but most small garages are better off letting AI draft the reply and suggest a slot while a person confirms it. Workshop time depends on the job, the ramp and parts availability, which the person on the desk judges better.
Further reads
- AI Service Plan Reminders for Heating and Cooling Customers — The same due-date logic for heating and cooling service plans.
- How to Set Up Missed Call Text Back for a Service Business — Catch the customers who ring back while you're under a car.
- How to Draft Customer Email Replies With AI That Sound Like You — Make AI-drafted replies sound like your front desk.
- How to Get More Google Reviews With Automated Review Requests — Ask returning customers for reviews without extra work.
- How to Use ChatGPT Scheduled Tasks for Business Reminders — A lightweight way to prompt yourself to run the weekly list.
- How to Segment Your Email List With AI, Even a Small One — Split customers into groups that need different messages.
- How Car Valeting and Detailing Firms Use AI to Fill the Diary — An empty-slot audit, rebooking intervals, photo-quote prompts and standby-list messages that help a valeting or detailing business keep its diary full.
- Winning Back Lapsed Dental Patients With AI Messages — How to find lapsed dental patients, split them by why they stopped coming, and use AI to write reactivation messages that don't guilt or scare.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.