Build Your Diesel Shop's CRM Around the Truck, Not the Customer

Median labor rate in heavy-duty shops hit $149 an hour by the end of 2025, up from $135 the year before, and more than half of shops still say they're understaffed. That's from the latest State of Heavy-Duty Repair survey, covered across the trade press this year. Two things follow: your hours are worth more than they've ever been, and you don't have enough of them.

Which is why it should bother you that a service writer can burn forty minutes rebuilding the history on one unit — a filing cabinet, a parts binder, and three text threads — before anybody picks up a wrench.

Build the CRM around the truck, not the customer

Every off-the-shelf CRM assumes a person buys something once. That is not your business. A fleet out of Findlay runs 38 trucks. Unit 214 has a service history that will outlast the fleet manager who calls you. The part you hung on it in March carries a 12-month vendor warranty nobody is tracking. The contact isn't the thing you're managing. The truck is.

So the core object in a custom CRM for a diesel shop is the unit: VIN, unit number, plate, engine family, current odometer, owning fleet, status. Repair orders hang off the unit. Parts lines hang off the RO, each carrying vendor, invoice number, warranty term, and the tech's photos. DOT paperwork — annual inspection certs, DVIRs you were handed, roadside reports — hangs off the unit too. One stack instead of three.

Then the pipeline gets built the way work actually moves through your bays: down and inbound, towed or mobile dispatched, diagnosed, estimate sent, awaiting approval, parts on order, in bay, road tested, invoiced, warranty window open.

That last stage is the one nobody builds. Most shops close the ticket the second it's invoiced. Leave it open and the system can watch it for you.

What the structure lets you automate

Structure first, automation second. Once the unit is the spine, this part gets easy:

• The downtime clock starts at first contact and every touch is stamped. When a fleet asks why 214 sat two days, you show them: call at 6:40 a.m., diagnosed 9:15, estimate out 9:50, approval came back 2:30 p.m.
• Warranty expiry is calculated the moment the part line is entered. Same unit, same component, inside the window? The CRM pulls the claim packet together — original RO, part invoice, fault codes, tech photos — before the core gets scrapped.
• Fleet accounts get a monthly record per unit: date, odometer, nature of the work, who performed it. Those are the fields federal rules require a carrier to keep for every vehicle, held as long as they have the truck plus six months. Hand them that and you're hard to replace.
• Estimate approvals are stamped against the version sent, which ends the argument about who authorized the aftertreatment job.

Where the AI actually sits

Not in the decisions. In the entry — which is where a shop bleeds.

A tech's two-minute voice note becomes a structured complaint, cause, and correction on the right repair order. A parts invoice PDF gets read into line items with the warranty term already set. A 4 a.m. text that says '214 down on 75 north of Beaverdam' gets matched to the unit, the account, and an open pipeline stage before your service manager finds the coffee.

Heavy Duty Trucking's reporting this year makes a useful point: AI in trucking is mostly showing up embedded inside existing maintenance systems rather than as standalone tools, and the real barriers are trust and data quality, not technical training. That's the case for specifying before building. AI sitting on a sloppy data model just produces confident nonsense faster.

So that's the order of work. A CRM spec — objects, fields, stages, who touches what and when. A data model that puts the unit at the center. A build plan phased so you get unit, RO, and parts working before anything else. Test it against two weeks of your own tickets, including the ugly ones. If it can't reconstruct the last comeback you ate, it isn't done.