Auto Repair CRM: Your Customer List Is Dirty

Somewhere in your system there are three customers named Mike Smith. One is “Mike Smith,” one is “mike smith,” and one is “MIKE SMITH DO NOT CALL.” All three drive the same F-150. Two of them are still getting oil change reminders.

Your auto repair CRM is your customer list, and your customer list is dirty. It is the most valuable asset in the building that nobody audits.

This post is about the record: what a CRM in a repair shop actually stores, where that record lives, how it breaks, and what to ask before you buy one. It is not about what to send people. If what you want is the campaign side, the reminder program, the review engine, the declined-work call script, the auto repair marketing playbook covers all of it, judged against gross profit per repair order, and this post will point you back there every time it gets close to telling you what to send.

I am here for the layer underneath. You can buy the best campaign tool in the industry, point it at a list where a third of the phone numbers are dead and half the vehicles have no mileage on file, and it will do exactly what you asked, faster, and wrong.

Read it in ten minutes. Audit your own list this afternoon.

Nobody sells you an auto repair CRM. They sell you one of two things

This is the paragraph missing from every vendor page on this topic, so here it is first.

In a repair shop, the customer record already exists. It was created the first time somebody wrote a ticket. It lives in your shop management system, next to the ROs, the parts, the labor, and the vehicle. Nobody built it on purpose. It just accumulated.

So when someone sells you a “CRM for auto repair shops,” they are selling you one of two things:

Archetype A: a feature set inside your SMS. The record stays where it already is. The tool adds the ability to act on it: texting, reminders, review requests, campaign segments. Tekmetric, Shop-Ware, AutoLeap and Mitchell 1 all sell some version of this.

Archetype B: a bolt-on that syncs a copy. An outside platform connects to your SMS, pulls a copy of your customer and vehicle list, and runs the communication from its own database. Steer, Kukui, Autoflow and AutoOps live here.

Naming which one you are buying is the entire decision. Archetype A means one database and one set of fields, with whatever limits your SMS already has. Archetype B means two databases that have to agree with each other forever, which brings us to the interesting part later.

Most auto repair shop CRM software falls into one of those buckets, and almost none of it says which.

One consequence: your options are mostly downstream of a decision you already made. In accounting firm Paar Melis’s 2025 benchmark report, built from client shops’ 2024 financials, the shop management system split across their client base ran Tekmetric 43%, Mitchell 21%, Protractor 11%, ShopWare 6% and RO Writer 6%. That is a client-base skew, not the whole market, but the point holds: whichever SMS you picked narrowed the CRM shortlist before you started shopping. If you are still choosing, read how to choose the best software for automotive repair first. That decision is the bigger one.

In a repair shop, the vehicle is the primary key, not the person

Here is why a generic CRM cannot do this job, and it has nothing to do with features.

Every general CRM in the world models a contact and a deal. A person, and a thing you are trying to sell them. That is a beautiful model for selling insurance.

A repair shop needs to model this instead:

• One customer.
• Three vehicles, two of them sold last year.
• VIN and plate on each.
• Mileage captured at every single visit.
• The service interval you recommended, per vehicle.
• Full service history, per vehicle, not per person.
• Declined operations, with a date and a dollar value.
• Which advisor owns the relationship.

Read that list again and find the sentence “the 2014 Odyssey at 148,000 miles that declined struts in April.” A repair-shop record holds it natively. A contact-and-deal record cannot hold it at all.

Could you build it in Salesforce? Sure. Custom objects, a vehicle-to-contact relationship, a mileage field, a declined-ops object. Congratulations, you now own a small software project and three bays, and nothing feeds it, because the parts, labor and history are all still in the SMS. So my honest recommendation on HubSpot, Salesforce and Zoho is: don’t. Good products, different problem. This is the rare case where “buy the vertical tool” is just correct.

What a CRM structurally cannot do, no matter what you pay

Worth saying plainly, because a lot of software gets bought to solve a problem it does not touch.

It cannot create data the front counter never captured. If mileage in is blank on the RO, no reminder engine can key off it. That record starts at the counter, which is why the auto repair order is the real beginning of your customer database.

It cannot create declined work nobody documented. If the inspection was verbal, there is no queue to work. Digital vehicle inspections are what turn “I told him about the struts” into a dated, priced record with a photo on it.

It cannot tell you whether the work was profitable. Your CRM watches the money coming in. Nothing in it knows whether the vendor ever credited the core on that customer’s last visit, which is a different discipline entirely: parts reconciliation.

Automating a dirty database just automates the mistake faster

This is my one strong opinion in this post, so let me back it with numbers instead of adjectives.

In the 2026 Ratchet+Wrench Industry Survey, self-reported responses from 430-plus shop owners and managers, only 15% of shops use marketing automation of any kind, while 20% of owners named marketing and customer communication as the biggest opportunity they see in AI.

Everyone reads that gap as a tooling gap. Shops know they should automate, shops have not automated, therefore sell shops automation.

I read it differently. Automation is a multiplier, and the thing it multiplies is your data. Point a clean engine at a dirty list and every error goes from occasional to industrial. The duplicate that used to cause one awkward phone call now sends two texts a month for a year. The 2004 Corolla that has been in a landfill since 2019 gets nine oil change reminders and never once complains, because landfills are famously bad at unsubscribing.

Now the math. This is illustrative, and the key assumption is mine, not a benchmark.

Take a shop with 3,000 customer records. I have no source for an industry duplicate rate, and I looked, so I am going to assume 8% and tell you plainly that it is my assumption. That is 240 phantom customers. Most are harmless. Say 30 have history split badly enough that the service record sits on one record while the current phone number sits on the other. The reminder never fires. The visit never happens.

Thirty repair orders. Using Paar Melis’s average repair order of about $702 at an overall gross margin of 52.3%, that is roughly $367 of gross profit each.

30 × $367 ≈ $11,000 a year, per shop. Across five stores, ≈ $55,000.

You cannot buy that back with a bigger subscription. There is no plan tier called Clean.

The seven-point audit: go look at your own list today

This is the part nobody sells, which is exactly why it is the part worth doing. Seven things to check, each with what it costs you. You need a spreadsheet, not a purchase order.

1. Duplicate customer records. Export the customer list, sort by last name, then sort again by phone number. The duplicates are adjacent and obvious once you look. Cost: reminders double-fire, service history splits in half, and your retention rate becomes fiction, because the same person counts as two customers and one of them looks like they never came back.

2. Vehicles with no mileage-in captured. Filter your vehicle list for a blank mileage field. Then look at how many visits recorded no mileage at all. Cost: mileage-based reminders are the tactic the marketing playbook rates highest, and they are literally impossible without this field. This is the field that makes them possible.

3. Dead numbers and landlines. Your list size is a vanity number. Deliverability is the real one. Pull the bounce and undeliverable report out of whatever tool sends your texts, and ask what percentage of sends actually land. Cost: a 4,000-record list where 900 numbers are dead is a 3,100-record list, and you are budgeting off the wrong number. Also, if you have never separated mobile from landline, some share of your “text-first” program is politely talking to a fax machine.

4. Missing or wrong VIN. Filter for blanks, then for anything that is not 17 characters. Cost: no VIN means no reliable year-make-model-engine, so the recommended interval is a guess and any parts or recall lookup keyed off it starts wrong.

5. Vehicles sold years ago, still generating reminders. Pull every vehicle with no visit in 36 months and no disposal flag. Cost: you are not annoying your customer. You are annoying a stranger who bought their car, and paying for the privilege. Mark them inactive.

6. Declined work stored with no dollar value or no date. Pull the declined and deferred operations report and count the blank amounts. Cost: this is precisely why nobody works the queue. You cannot sort a list by value when half the values are missing, and an unsortable list is an archive.

7. Opt-out and consent status scattered across tools. If a customer replies STOP to your texting tool, does your SMS know? Does the email platform? Does store three? Cost: this is the one with actual regulatory teeth, so understand it rather than guess. Application-to-person texting runs through carrier registration now, and The Campaign Registry, which operates the 10DLC ecosystem, describes it as a channel in which brands and campaign service providers “are verified prior to being allowed to send messages.” Your provider handles registration. What no provider can fix is a consent status that exists in four systems and agrees in none of them.

Notice what all seven have in common. Not one is fixed by buying software. Every one is fixed by an afternoon, a spreadsheet, and somebody who cares.

The reason to bother, from the same Ratchet+Wrench survey: 86% of shops track KPIs, and the ones that do are three times more likely to clear $1 million in revenue. Retention is a KPI. You cannot track it on a list where the same customer appears twice.

The seam between your CRM and your SMS is where the record breaks

Buy Archetype B and you now have two databases holding the same customer. That is not a flaw. It is a fact you have to manage, and three questions decide how badly it goes.

Which direction does the sync run? One-way out of the SMS is the common default. Your CRM learns about new customers and tells the SMS nothing back. So when a customer updates her number by replying to a text, does the front counter keep dialing the old one forever? Somebody has to win that conflict, and you should know who before it matters.

Does a booking write back? This is the biggest one. A customer books online at 9pm. Does that land on the schedule your advisor opens at 7am, or in a queue somebody has to re-key? A booking tool that does not write into the SMS has not saved anyone any work. It has moved the typing.

Which fields cross the seam? Name, phone and email always make it. Mileage in, VIN, declined operations with amounts and dates, and advisor of record are the ones that separate a real repair-shop CRM from a mailing list with a service logo on it.

What to look for in auto repair CRM software

I will not rank tools for you, because the right answer depends on which SMS you run, and that is not a ranking, it is a constraint. Judge any auto repair shop CRM on four things: does it own the record or hold a copy, which fields does it read, does it write back, and is the database per location or per group. Everything else is a feature list.

The questions to ask the salesperson

Copy these. Ask them in this order. Take notes, because the answers to numbers 5 and 8 are where demos get quiet.

1. Does your product own the customer record, or hold a copy of what my SMS has?
2. Is the sync one-way or two-way, and how often does it run?
3. When your database and my SMS disagree about a phone number, which one wins?
4. Which fields do you actually read? Specifically: mileage in, VIN, declined operations with dollar values and dates, and advisor of record.
5. When a customer books online, does the appointment land on my SMS schedule automatically, or does somebody re-enter it?
6. Is the customer database one per location or one per group? Can store four see store two’s history?
7. Is the opt-out list shared across all my locations, or one per store?
8. Do you dedupe on import? On which field? Show me it running on my actual export.
9. If I cancel, what data do I get back, in what format, and how long does it take?
10. Is pricing per user, per location or per group, and what does it cost when I open store six?

Number 8 is my favorite. Ask them to import your real export, dirt included, and show you what their tool does with three Mike Smiths. A vendor who says yes to that is worth talking to.

Auto repair CRM vendors, and which archetype each one is

Nine tools worth knowing. Everything in this table came off the vendor’s own site while I was writing this, and where a vendor does not publish a price, the table says so instead of guessing. I am not ranking them, because your SMS already did most of the ranking.

Two things that table taught me that I did not expect.

This category has been consolidating, and the table hides it. Shopgenie is now Tekmetric Marketing. AutoOps and Mechanic Advisor are both Steer properties. So two of the nine rows above are the same company, and one is a product that no longer exists under the name most owners still search for. I’m not printing deal dates, because none of these companies dates the transaction on its own site. Check who owns whom this quarter instead. The org chart moves faster than the search results do.

Almost nobody publishes a price. Three of the nine do, and two of those three (Tekmetric and AutoLeap) publish the identical $179, $309 and $409 on annual billing. That is not vendors being shifty, it is a category where the answer depends on store count and what you bundle. It does mean any price you read on a comparison site is somebody’s guess, including a guess about your shop.

For a wider view of the tool classes around this one, our rundown of automotive software solutions maps the marketing and CRM category against the rest, and the six-layer repair shop software stack shows where CRM sits relative to your ledger and your AP.

Five stores, one customer, five records

This is where the record question stops being tidy and starts costing real money.

A customer moves across town. She has used store two for six years. Today she pulls into store four. What does store four see?

If the answer is “nothing,” you just paid full acquisition price for a customer you already had, and your advisor is about to recommend a service she declined in April at a different building with your name on it.

Four decisions a group has to make on purpose.

Shared database or siloed per store. Both are defensible. Siloed is simpler and keeps store-level reporting clean. Shared means store four sees the Odyssey at 148,000 miles and the declined struts, which is the entire reason you own five stores instead of five businesses. Pick one deliberately. What you do not want is the accidental third option: shared-ish, where some stores were migrated and some were not, and nobody can tell you which.

Who owns the relationship for reporting. If she used store two for six years and store four once, whose customer is she? Whose retention number does she land in? There is no universally right answer, but there is a wrong one: not deciding, then discovering at year end that two general managers have both counted her.

The opt-out trap, and this is the expensive one. Consolidating the customer database without consolidating the opt-out list is worse than not consolidating at all. Store one’s unsubscribe does not stop store four’s text, so the same person ends up telling five stores to stop, one at a time, getting angrier each time. If you merge the records, merge the consent list in the same project. Not in phase two.

You cannot compare retention across stores until the records are deduped. This is the one I would put on the wall. Store three’s retention looks eleven points worse than store one’s. Advisor problem, market problem, or does store three simply have twice as many duplicates, so every returning customer looks new? Until you know, that dashboard is not management information. It is management fiction, arrived at with real effort and a nice chart.

One pricing note that only bites at scale: five stores of Tekmetric’s Marketing add-on is $20,700 a year before the plans underneath it, which is the arithmetic that never happens in the demo. Ask question 10.

Review velocity per store is a genuinely great group metric, and it belongs to the marketing playbook’s multi-location section, not here. Same for how you actually run the reminders. If you are working out how much of the back office to centralize while you sort this out, running a multi-location back office is the companion piece.

Clean the record, then buy the tool

Here is the whole post in five lines.

1. Find out whether your customer record lives in your SMS or in a copy. That is the decision, and most shops have never named it.
2. Run the seven-point audit. Spreadsheet, one afternoon, no purchase order.
3. Fix mileage in and declined-work dollar values first. They are the two fields everything downstream depends on.
4. Only then shop. Take the ten questions with you and ask number 8 out loud.
5. If you run more than one store, decide shared or siloed, and merge the opt-out list on the same day you merge the records.

Almost every shop I talk to about this thinks they have a software problem. Nine times out of ten they have a data problem wearing a software problem’s clothes, and the tool they are about to buy will work perfectly, which is the bad news.

Trust, but verify. Then go look up how many Mike Smiths you have. If the answer is three, at least one of them owes you for an oil change.