Guide
How to automate a repair shop, in the order that actually works.
Most shops automate in the order their vendors call them. A presence product, then a review tool, then an ad campaign, and eighteen months later the phone still rings out at lunch. The order matters more than the software, and there is a sequence where each stage makes the next one cheaper.
The governing principle
Every automation either protects work you already have or buys work you do not.
Protecting existing work is arithmetic that works in your favor: the vehicle is already known, the customer already chose you, and the remaining obstacle is attention rather than demand. Buying new work only pays off if everything downstream of it holds.
That gives the ordering rule. Fix capture before nurture, because a reminder sent to a customer who could not reach you last week is worse than no message. Fix the work you already quoted before you buy leads, because an unanswered estimate costs nothing to chase.
The test for any proposed automation: if it stopped working tomorrow, how long would it take you to notice? If the honest answer is weeks, it is not protecting anything yet. Put it later in the sequence.
The plan
Five stages. Do not start the next one until the current one holds.
Each stage has a gate: a thing that has to be visible in your own numbers before the next stage is worth starting. The gates are deliberately unglamorous.
Nothing in this sequence requires a purchase. A shop can run stages zero through two with a phone log, a list, and somebody willing to make the calls. That is the point of the ordering: the cheapest work happens first, and it pays for what comes later. See the five leaks this sequence closes ->
The other list
Do not automate these yet.
Not never. Not at this stage. Each of these is either a judgment call, a pricing decision, or a customer-facing moment where a mistake costs more than the time it saves.
Honest limits
Three problems that look like automation problems and are not.
A shop nobody wants to call. If the reviews describe long waits and surprise charges, more messages will reach more people and produce the same conclusion. Fix the operating problem the reviews describe first.
No capacity to absorb the work. A shop with one empty bay and a two-week backlog does not need more leads. It needs a scheduling conversation, which is a different exercise. See capacity planning.
A team that does not trust it. An advisor who believes the system will annoy customers will quietly stop approving messages, and the process dies without anyone deciding to kill it. Start narrow and expand when they ask for more.
The useful mental model is that automation removes the excuse. Once the calls are answered and the estimates are chased, the remaining constraint is visible: usually capacity, staffing, or pricing. That is a more honest place to be, even when the answer is uncomfortable. See how the layers fit together ->
If you only do one thing
Stage one, this week, without buying anything.
The measurement stage is free and it decides whether any of the rest is worth your money.
Export last month's inbound call log. Mark every call that was not answered during business hours.
Open the form inbox, the chat tool, and the profile messages. Time how long the oldest unanswered one has sat there.
List every open estimate. Sort into three piles: waiting on the customer, waiting on the shop, and nobody knows.
Count the approved jobs with no appointment. This is the pile most owners have never looked at.
Multiply each pile by your average repair order and the share you think would say yes. Pick the largest. That is stage one.
Start with the pile you have not counted
Bring last month's numbers. We will pick the first stage with you.
Missed calls, open estimates, approved jobs with no date. We will tell you which one is worth automating first, and which ones to leave alone for now.