How to Run a Repair Shop: The Complete Guide

Whatever you fail to write down at intake, you pay for later. The exact condition of the device, the unlock code or pattern, the accessory left in the box, the fault described in the customer's own words rather than yours. Those five minutes at the counter save hours of argument three weeks later. They also protect you: a scratch you did not record is a scratch the customer will sincerely believe you caused.

One rule changes more than any other. A device taken in without a unique identifier is a device you will eventually lose. Serial number, IMEI, or a printed label stuck on the case — the medium matters less than the fact that there is one, and that it can be scanned. A technician who searches visually for one black phone among forty burns far more time than anyone in the shop believes, and the search happens several times a day.

Photograph the device at intake. Two shots, front and back, ten seconds. It is the cheapest insurance in the trade. Arguments about pre-existing damage end the moment you can show a timestamped photo taken before anyone opened anything, and they end without anyone raising their voice. Keep the photos attached to the job, not in a phone gallery that only one person can reach.

Set expectations at the counter, not by message two days later. Give a range instead of a promise: two to four working days if the part is on the shelf, up to eight if it has to be ordered. A customer told eight days who gets the device back in five is delighted. A customer promised two days who gets it in five is unhappy — even though that repair was objectively faster. The turnaround did not change; the expectation did.

A quote accepted verbally is worth nothing on the day the customer disputes the amount. That is not cynicism, it is simply what happens when a repair costs more than someone remembers agreeing to. Written approval — a signature on a tablet, a reply to a link, a short email — takes seconds and ends the conversation before it starts. Ask for it every single time, including for the regulars, especially for the regulars.

Quote in at least two lines: parts and labour. A customer who sees one number compares it to the part price they found online and concludes you are expensive. A customer who sees the part, the labour and the warranty period understands what they are buying. Same total, completely different reading. Adding the warranty line is the cheapest objection handler you will ever write.

Charge for diagnosis and say so up front. Free diagnosis sounds generous and quietly trains people to use your bench as a free second opinion before they buy a new device anyway. A modest diagnostic fee, deducted from the repair if the customer goes ahead, costs you no real jobs and removes a surprising amount of unpaid bench time. Shops that introduce it usually see the volume of quotes fall and the acceptance rate climb.

When a job turns out worse than expected — a second fault, a board-level problem, a screen that arrives with a dead line — stop and re-quote before you continue. Absorbing the difference silently to keep the peace teaches you that your own estimates do not matter, and it burns margin on bench time you already committed. A short honest message explaining what you found is accepted far more often than shops expect.

Count your incoming calls for one week and write down what each one is about. In most shops a large share come down to a single question: is it ready? Each call costs two or three minutes, plus the time to find the device, plus the real cost — picking the interrupted job back up. The interruption is the expensive part, not the conversation.

The fix is not to answer faster. It is to remove the reason for the call. An automatic message on every status change — received, quote ready, in repair, waiting for parts, ready for collection — deletes the question. The customer knows, so the customer does not call. Shops usually notice the effect within about two weeks, first in the afternoons, which is when repair work and phone traffic collide hardest.

The side effect is commercial rather than organisational. An informed customer collects sooner, and a collected device frees a shelf slot and triggers payment. Most shops that switch on automatic updates see the collection shelf empty before they notice the phone going quiet, and the two effects compound: fewer devices on the shelf means less searching, which means fewer interruptions.

For any of this to work you need one source of truth for status. If the status lives in a technician's head, or on a sticky note on the bench, there is nothing for the automation to fire from. This is where software earns its place: the status change you record for yourself is the same event that notifies the customer, with no second action to remember.

Two symmetric mistakes destroy margin. Too much stock means cash frozen in screens for models nobody brings in any more — and screens do not improve with age, they get harder to sell every quarter. Too little stock means repairs waiting on a delivery, devices sleeping on a shelf, and customers calling to ask why. Both mistakes are usually made at the same time, on different references.

The useful measure is not how many parts you hold, it is how fast each reference turns. A part that goes out four times a month deserves a permanent shelf slot. A part that went out once in six months is ordered per job, even when the unit price stings. Sorting your references by movements over the last twelve months takes an afternoon and almost always shows that a small minority of them carries most of the volume.

Link every part to the repair that consumed it. Without that link you will never know what a job actually earned. A screen replacement at an attractive headline price can lose money once you count the part, the adhesive, the protective film, the ten minutes of testing and the one job in twenty that comes back under warranty. The averages only become visible when the consumption is recorded job by job.

Lead time matters as much as unit price. A supplier who is three euros cheaper and four days slower costs you a shelf slot, a chasing customer and sometimes the job itself. Keep a second source for your top twenty references, and record what each supplier actually delivers rather than what the website promises. The full method is in our guide to spare parts inventory management.

Every shop has its shelf of shame: repaired devices, paid or unpaid, that nobody ever came to collect. They take space, they lock up parts you could have used, and after a while they raise a legal question you would rather not have to answer. The pile grows slowly enough that nobody notices until it takes a whole shelf.

The treatment is simple but has to be systematic: a reminder at fifteen days, a second at one month, a formal written notice after that. What stops shops is never the procedure, it is having to remember to run it while doing everything else. Automate the reminders and the problem leaves your head entirely. Keep the written trace of each reminder — it is what makes the later steps defensible.

On payment, take the money at collection, not afterwards. It sounds obvious, and the failure mode is always the same: the friendly regular who will settle up tomorrow. Deposits on ordered parts solve most of it. When a part is model-specific and cannot be returned, asking for a deposit is not aggressive, it is the normal way of not financing someone else's change of mind.

Card payment at the counter should not live in a separate terminal with a separate ledger. When checkout is part of the repair record, closing the day takes minutes. When it is not, somebody spends the end of every week matching bank lines to jobs from memory. The same logic applies to device buy-back: money going out of the till needs a record as complete as money coming in.

Average turnaround from intake to collection tells you whether you keep your promises. Average ticket per repair tells you whether your pricing holds. The share of devices uncollected past thirty days tells you whether your follow-up works. The return rate within warranty tells you about part quality and workmanship, and it is the number most shops avoid looking at. Bench hours billed against bench hours worked tells you where the time leaks.

Everything else is decoration. A dashboard with forty indicators is a dashboard nobody opens. Pick those five, look at them on the same day every month, and act on the worst one before touching the others. A shop that fixes one number per month is far ahead of a shop that measures thirty and changes nothing.

Staffing is the quiet variable behind all five. If you do not know who was on the bench when, you cannot explain a bad turnaround month or a good one, and you cannot cost a job honestly. Recorded hours turn arguments about effort into arithmetic, which is a much shorter conversation.

A word on tools, then. SERVO grew out of Maison Du Geek, a working repair shop in the south of France, and it covers this whole loop: intake with printed labels, status tracking with automatic SMS, quotes and invoices, parts and suppliers, checkout, device buy-back, staff time tracking and multiple locations — €49 per month excluding VAT, with a 30-day trial and no card required. The tool matters less than the principle: the loop described above has to live somewhere other than your memory.