How to Build Training Material With the Tools You Have

Why this matters

Shops do not skip building training material because they think it is a bad idea. They skip it because the mental picture is a studio, an editor, and a course platform, and none of those are happening this quarter. The picture is wrong. Everything you need is already being generated by the work: job photos you are required to take anyway, a callback log that names your real gaps, and the rescue phone calls where one person explains something to another. The job is not authoring. It is harvesting, and harvesting costs minutes, not budget.

Step 1: Harvest, do not author

Sit down to write training material from scratch and you will produce generic content, slowly, that reads like it came from anywhere. Every good asset a shop owns is a byproduct of real work that was captured instead of discarded. Five sources produce almost all of it:

• The callback and rework log. This is your curriculum and it is already written. Every callback is the shop telling you, in its own handwriting, which step people get wrong. Nothing you invent will be better targeted.
• The rescue call. Any time one tech has to phone another to proceed, the answer given on that call is a finished lesson. Record it, with the parties knowing it is being recorded.
• Job photos. You are already shooting before, during, and after. A handful of those, labeled, become a "what right looks like" reference at no incremental cost.
• Notes where a tech explained something to a customer. A tech who has written a clear explanation of a failure into a job note has already produced teaching prose in their own voice, which lands better with the crew than anything management writes.
• The estimate that was badly wrong. The gap between quoted and actual hours on a job type is a lesson about scoping that no amount of classroom time transfers.

If you skip this step, you write from imagination, and the material addresses problems your shop does not have while missing the one that generated six callbacks last quarter.

Step 2: Pick the format by what the learner will be doing when they need it

Do not pick by what is easiest to make. Pick by the posture of the person who will consume it.

The recurring mistake is filming something that should have been a checklist. Video is unscannable. A tech standing at a unit with cold hands cannot scrub a six-minute video to find step four, and after one attempt they will stop opening your videos entirely.

Step 3: Shoot to a three-minute standard, and never edit

Set a hard ceiling of about three minutes on any single video and a strong preference for ninety seconds. One take. No intro, no title card, no editing. Editing is where this program dies, because it converts a five-minute task into an evening.

Three non-negotiables, and only three:

• Audio has to be audible. Wind and running equipment kill more shop videos than anything else. Get close, or narrate afterward in the truck over the footage while it is fresh.
• The thing has to be in frame. Sounds obvious. The most common defect is a video of someone's shoulder.
• The content has to be correct and safe.

On that last one: if the capture shows work on gas, water combined with electricity, stored energy, height, or a pressurized system, the safety step must be visible in the shot or explicitly narrated. A video that opens with the panel already off and no mention of lockout teaches that lockout is optional, and it teaches it with the authority of your best tech. If the safety step was skipped during the capture, throw the capture away and re-shoot. Publishing it because the rest of the content was good is how a shortcut becomes shop standard.

Say the conclusion out loud at the end of every video: what the viewer should now do differently. A video that ends when the task ends leaves the lesson unstated.

Step 4: Convert one job into three artifacts

The efficient unit of production is not one asset. It is one job, harvested three ways, because the setup cost is already paid once you are standing there.

Take a job type that generates callbacks. On the next occurrence, with the tech's knowledge:

1. A ninety-second narrated video of the single step people get wrong. Not the whole job. The one step. Shot on the phone already in their pocket.
2. Six labeled photos of what correct looks like at the checkpoints where it can go wrong. Four of those six are shots the documentation standard already requires, so the incremental cost is two extra photos and the labeling.
3. A one-page pre-flight checklist for that job type, derived from the callback log rather than from memory. This one is desk work, done once.

Field time: roughly five minutes for the video including setup, and a couple of minutes for the extra photos and labels. Call it seven incremental minutes on a job that was going out anyway. Desk time: about twenty minutes to build the checklist from the log, plus a few minutes to caption the photos. Total, about half an hour of shop time to produce three assets that get used on every subsequent occurrence of that job type.

The payback, worked through. Say your log shows this job type ran 40 times in the last year with a 15% callback rate, which is 6 callbacks. Time a callback on this type honestly: drive out, diagnose, correct, drive back, with no revenue attached. Suppose it runs 2.5 hours. That is 15 hours a year of unpaid rework on one job type.

Now suppose the checklist and the photo set bring the callback rate down to 8% over the following year. At 40 occurrences that is 3.2 callbacks instead of 6, so 2.8 avoided, at 2.5 hours each, which is 7 hours recovered against about 0.5 hours invested. Roughly 14x in the first year, and the assets keep working in year two for nothing.

Use your own two numbers here rather than these, because the whole point is that both are already sitting in your records: the occurrence count and the callback rate for that job type. If you do not know your callback rate by job type, that is the first thing to fix, because without it you cannot tell which job type to harvest.

What changes this calculation. If a job type runs 4 times a year rather than 40, the payback stretches to several years and the honest move is a written entry rather than a photo and video set. If the callback cause is a parts or supplier problem rather than a technique problem, no training asset will move the rate and you have misdiagnosed the gap. Check the callback causes before you produce, not after.

Step 5: Write the one-page format so it survives a phone screen

Whatever you produce in writing lives on a phone, held in one hand, often in poor light. That constrains the format more than the content:

• One page. If it needs two, it is two documents.
• Short lines. A paragraph of six lines on a desktop is a wall on a phone.
• Bold the action word at the front of each step so the eye finds its place after looking away at the equipment.
• Values and thresholds in a table, never buried in prose, because that is what gets looked up mid-task.
• Name the starting condition at the top: powered down, isolated, panel removed, customer present.

Step 6: Leave the raw material raw

There is a strong pull to polish a rough capture before releasing it. Resist it. A shaky ninety-second phone video that exists and gets watched beats a clean one that is still on the to-do list, and every shop that has tried to polish has ended up with a small library and a large backlog.

The exception is anything that will be seen outside the shop, or anything tied to a documented compliance procedure. Compliance material has to match the official procedure exactly, or you have created a second, conflicting authority, which is worse than having none.

Also check every capture for two things before it goes into the library: no identifiable customer information or property markers in frame, and no person in it who has not agreed to be filmed.

Step 7: Put it where the work is

An asset in a shared folder is an asset that only gets used by people who already know it exists. Attach each one to the thing it explains: the job template, the equipment record, the checklist step, the service line. If a tech opening a job type sees the video and the checklist without searching, usage goes up by a lot. If they have to remember and then navigate, it does not.

Step 8: Run a cadence that is self-selecting

The sustainable rule, and the one worth putting on the wall: every callback produces one artifact.

It works because it is self-limiting and self-prioritizing. A good week generates no homework. A bad week generates exactly as much curriculum as the shop just paid for in rework, aimed precisely at what went wrong, and the person who produces it is the person who just lived it. Nobody has to decide what to make next, which is the decision that stalls every content plan.

Add one rule so it does not degrade: the artifact is produced within the same week as the callback, by whoever handled it, and it is short. A callback that generates a promise of a thorough writeup produces nothing.

How to verify the material actually works

• Cold test each asset once. Hand it to someone who was not on the job and have them narrate what they would do. Every question they ask is a defect in the asset, not in the reader. First drafts typically generate three to six questions and most are unexplained shorthand.
• Watch the callback rate on the harvested job type. This is the only outcome measure that matters, and it is why you harvest from the callback log rather than from wherever seems interesting. If the rate does not move over the next batch of occurrences, either the asset is not being used or the cause was never technique.
• Check usage, not production. A library that grows while nothing gets opened means the assets are not in the flow of work. That is a placement problem, addressed in step 7, not a content problem.
• Re-watch your oldest videos once a year. Material ages quietly: a supplier process changes, a code cycle updates, a technique improves. Wrong material carries your shop's authority, so retire it rather than leaving it there looking current.

References

• OSHA general industry standards on training documentation and hazard-specific procedures
• U.S. Small Business Administration (SBA), guidance on low-cost workforce development
• See related: Checklist vs SOP vs Training: A Decision Matrix
• See related: How to Use Photos and Video as a Teaching Tool
• See related: How to Document Tribal Knowledge So It Outlives the Person