How to capture a retiring diesel mechanic's troubleshooting knowledge

Why the final two weeks are the wrong place to start

Most yards begin this at the exit interview. NASA studied its own version of the problem and found the same reflex. Its 2021 report on knowledge continuity notes that capture tends to be "retroactive after identifying the individual who is departing," and argues that "there should be knowledge capture throughout the career, not after someone has given notice." A survey respondent in that report put the difficulty plainly: "the specific experience is hard to capture in a policy memo."

The arithmetic is against you as well. The Bureau of Labor Statistics counted 319,900 diesel service technicians and mechanics in 2024, projected 2 percent growth from 2024 to 2034, and estimated about 26,500 openings each year over that decade, most of them replacements for people leaving the trade. BLS classifies the path into the job as long term on the job training.

So the person filling the seat needs years of exposure and the person leaving has weeks. That asymmetry is the whole problem, and no amount of effort in the last fortnight closes it. What you can do is decide, early, which slice of an arcane career is worth the recording time.

The ninety day sequence

Four stages, each with a different job. Only one of them involves recording.

What to ask, engine family by engine family

Generic questions get generic answers. Sort the list by what is actually parked in your yard, then ask about each family in its own parlance. Cummins, Detroit, Caterpillar, PACCAR, International, whatever the mix happens to be.

Then go system by system. Aftertreatment and regen behaviour on short duty cycles. Air systems on cold starts. Charging and grounding faults that present as something else entirely. Harness routes that chafe on one particular chassis year. Coolant and DEF quality problems that read on the dash as a sensor failure.

Ask about the diagnostic order, not just the repair. How does this technician decide what to check first? That heuristic is the most perishable thing in the shop and the least likely to be written anywhere.

The same logic applies to intervals. A factory schedule assumes a duty cycle your routes may not resemble, which is the subject of our piece on why generic maintenance intervals fail without local context. The retiring technician has been quietly correcting for that gap for years. Ask them how.

How long each recording should run

Short. NASA's knowledge continuity guidance for supervisors recommends clips of three to five minutes, on the grounds that they can pass information without overwhelming the viewer. The number holds up in a shop for a second reason. A technician hunting an answer between work orders will not scrub through forty minutes of footage to find ninety seconds of substance.

Audio is usually enough. A phone propped on a fender captures a diagnostic explanation without ceremony. Reach for video only when the knowledge is spatial: a hand position, a routing path, a sequence of movements that words make worse. A chest mounted camera such as a GoPro works, and so does a phone on a tripod. If the knowledge lives inside diagnostic software rather than under the hood, a screen recorder such as Loom captures the click path better than any camera will.

The equipment is never the constraint. The constraint is whether anybody can find the clip afterwards.

The records the rule requires, and the ones it does not

49 CFR 396.3 requires a motor carrier to systematically inspect, repair and maintain its vehicles, and to keep records identifying each unit, the nature and due date of the inspections and maintenance, and the dates and nature of the work performed. Those records stay where the vehicle is housed or maintained for one year, plus six months after it leaves the carrier's control. Our federal inspection rules reference lays the requirements out in full.

Read that list again and notice the lacuna. The rule wants to know what was done. It has nothing to say about why this technician checked the ground strap before the sensor, or which supplier still stocks the sending unit for a 2014 chassis. The compliance file is a ledger. It is not an explanation.

A yard can be fully compliant and still lose everything useful on a retirement date. The gap between the ledger and the reasoning is exactly what you are documenting.

What to do with the files so they still work in three years

The common failure is not a dearth of recordings. It is a folder of recordings nobody can search. Forty hours of unindexed video is a shipwreck: it exists, and nothing is recoverable from it.

Transcribe within a week, while the session is fresh enough for the veteran to correct. Then tag each clip three ways. By unit or engine family. By system. And by the fault code or symptom a technician would type at the moment they hit the problem. That third tag is the one that matters, because nobody searches for a veteran's name and the word interview. They search for the code on the dash.

Have somebody review it before it goes live. A retiring technician's workaround can be shrewd, or it can be a stopgap from 2009 that nobody revisited. Approval by the foreman keeps the corpus honest and puts the veteran's name on the answer their colleagues end up using. Our shop knowledge page covers how that review step works in practice.

Where YardWise fits

The method above works without software. Recordings, transcripts, a shared drive and some discipline about tagging will carry you most of the way, and a yard that does only that is far ahead of one that does nothing.

What breaks is retrieval three years later, when the person who did the tagging has also gone. YardWise is the place the answers land. A technician holds the button, asks in plain speech, and hears the answer read back with two citations: the manual page it came from, and the name of the person in your yard whose note was approved. Capture runs the same way, a few minutes between work orders rather than a writing project.