How to Train the Trainer in a Shop of Five

Why this matters

In a shop of five there is no trainer. There is your best producer, who is also the person you hand every new person to, and every hour they spend teaching is an hour of your most valuable output. So teaching gets done in the cracks, badly, by someone who was never taught how, and the shop concludes that training does not work here. It is not a motivation problem and it is not a talent problem. Teaching in the field is a small, learnable skill set, maybe four moves, and a person who has the four moves teaches better in twenty minutes than an untrained expert does in a day.

Step 1: Pick on teaching behavior, not on skill rank

The reflex is to hand training to whoever is technically strongest. That is the wrong sort. Watch for three behaviors and pick on those:

• Do they let someone struggle? The single strongest predictor. A person who takes the tool at the first hesitation cannot be a trainer until that habit breaks, no matter how good they are.
• Do they ask before they tell? Someone who naturally says "what do you think it is" is already doing the most valuable move in field teaching.
• Can they be wrong out loud? A trainer who has to be right in front of a learner will hide the reasoning, and the reasoning is the whole product.

The best tech in the shop may score badly on all three. That does not disqualify them from the work, it means they need step 2 more than anyone. But if you have a second-strongest tech who scores well, start there and you will get further faster.

Step 2: Teach exactly four moves, and nothing else

Resist building a curriculum. Four moves is the entire trainer syllabus, and you can teach all four in one 90-minute sitting plus one observed job.

Move 1: narrate the decision, not the motion. Learners can see the hands. They cannot see the choice. "I'm checking this first because it is cheap and reversible, and if I'm wrong I've lost two minutes instead of an hour" teaches. "Now I check this" does not. The trainer's job is to speak the part of the work that is invisible.

Move 2: ask before you tell. At every decision point, the trainer asks what the learner would do next and why, and waits. Three to five seconds feels unbearably long and is where the learning happens. When the answer is wrong, the trainer says why it is wrong before saying what is right, because the why generalizes and the what does not.

Move 3: one correction at a time, at the right moment. A learner receiving four corrections retains roughly none, and feels supervised rather than taught. Pick the one that matters most and let the other three go until later. Timing matters as much as count: correct safety immediately and out loud, correct technique at the next natural pause, and correct customer-facing behavior after the customer is gone, never in front of them.

Move 4: close with teach-back. At the end, the learner explains the job back: what it was, how it was diagnosed, what they would do differently. Two or three minutes. This is where the trainer finds out what actually landed, and it is routinely nothing like what they assumed. See related: How to Have a Trainee Teach It Back to You.

That is the whole syllabus. A trainer who does these four things consistently will out-teach a certified instructor who does not.

Step 3: Give them the words for the two hard conversations

Both of these conversations are why people avoid the trainer role, and both are fixed with a script rather than with courage.

The correction that does not deflate. Three parts: name the behavior, name the consequence, ask for the change. "You went to the meter before isolating. If that circuit had been live, that is a hospital trip. Isolate and verify first, every time, even when you are sure." No preamble, no compliment sandwich, no lecture. Then move on and do not bring it up again that day.

The not-yet sign-off. The harder one, because the trainer has to tell a colleague they are not cleared. Script: name the specific objective, name the evidence, name the next step and the date. "You are clean on everything except the diagnostic sequence, I saw it go out of order twice today, so that stays open. Two more paired jobs on that type and we re-check on the 14th." Specific, bounded, dated. Vague disappointment is what makes this conversation dreaded; a named gap with a date attached is just work.

Have the trainer say both of these out loud to you in the shop before they need them. The first time should not be live.

Step 4: Protect the clock with an explicit, stated trade

This is the step small shops skip, and skipping it kills the program every time. Teaching costs output. If you do not name the cost and absorb it, the trainer will absorb it by teaching less, and they will be right to.

Say the number out loud to the whole crew: on a teaching day, this person's expected output is roughly two-thirds of normal, and that is the plan, not a shortfall. Then follow through in the two places it actually gets tested:

• Dispatch. The pairing does not get split for coverage on a busy Wednesday without your approval. If it can be split whenever things get tight, it will be split every week.
• Any performance number the trainer is measured on. If the trainer is on a bonus, an efficiency metric, or a billable-hours target, adjust it for teaching days. Asking someone to teach while their own scorecard punishes them for it is asking them to choose, and they will choose the scorecard.

Step 5: Watch one job and coach the coach

Ride along on one full job with the trainer and the learner. Say nothing. Take notes against the four moves only:

Debrief with the trainer alone, and use the same rules on them: one correction, named behavior, named change. Coaching the coach with four criticisms teaches them exactly what not to do.

Step 6: Build the second trainer before the first one burns

One trainer is a single point of failure and a burnout risk. The person who trains everyone is having the same conversation for the fourth time, on top of a full workload, and it wears out quietly.

The cheapest second trainer is the person who was just trained. Someone six months past their own shadowing week remembers exactly which parts were confusing, which is a real advantage over a twenty-year tech who has forgotten what hard felt like. Give them a narrow scope: they train only on the job types they are cleared on, and only the four moves.

Worked example: a five-person shop over one quarter

Five people: owner, two senior techs, one mid-level, one apprentice. Historically the owner did all the teaching, at maybe three hours a week squeezed around estimates, and the apprentice's progress had stalled for two months.

Selection. The technically strongest tech takes the tool at the first hesitation, consistently. The second-strongest asks questions naturally and is comfortable saying they are not sure. The owner picks the second tech, and tells the first tech directly why, framed as a habit to break rather than a judgment.

Setup. One 90-minute sitting on the four moves and the two scripts. The owner rides one job in week 2 and counts: 11 decision points on the job, 4 of which the trainer explained the reasoning for, 2 questions asked with real pauses, 6 corrections issued (four more than useful), and no teach-back because the day ran late.

One correction, applied to the trainer. The owner names only the correction count: pick one, let the rest go. Nothing about the missed teach-back, nothing about the narration rate, both of which were also weak. Two weeks later, the second observed job shows 3 corrections instead of 6, and the narration rate had risen on its own to 7 of 12 decision points without anyone raising it, because slowing the correction habit created room to explain.

Cost. Roughly 8 paired jobs per month at about a third of a day of output given up per job, so on the order of 2.5 to 3 days of production per month across a five-person shop. That is close to 3 percent of the shop's monthly field capacity.

Return by end of quarter. The apprentice, stalled for two months, cleared three job types for solo dispatch. Dispatch went from four schedulable names on those types to five, which on the shop's most common work is a 25 percent increase in schedulable capacity. The owner's own teaching time dropped from about 3 hours a week to under 1, and the hour that remained was spent on the trainer rather than on the apprentice, which is the leverage the whole exercise was after.

The honest part: the first month showed almost nothing. Output was down, the apprentice was not visibly better, and it looked like a mistake. That is normal and it is why the trade in step 4 has to be stated in advance. A shop that judges this at four weeks kills it right before it pays.

What changes the answer

A shop of two. There is no selecting, and there is no protecting a clock that does not exist. Compress to move 1 and move 4 only, applied to whatever job you are already doing together. Narrate the reasoning, ask for a teach-back at the end. That is achievable on a normal day and it is most of the value.

A trainer who is genuinely bad at it after real coaching. Some strong techs will not get there, and forcing it produces a bad experience for the learner and resentment in the trainer. Reassign them to the knowledge-capture role instead: they are a superb source for procedures and error catalogs even when they cannot teach live. See related: The Tech Who Can Do It But Cannot Explain It.

Heavy seasonal load. Do not launch this in your peak. Start in the slow season when the output trade is cheap, so the habits are set before the pressure arrives.

Multiple learners at once. Beyond one learner per trainer, quality falls off fast in field teaching because the moves depend on individual attention. Stagger the learners rather than pairing two to one trainer.

How to verify you got this right

• Ride a job and count the four moves. Numbers, not impressions. A trainer who has the moves shows them in the counts, and a trainer who does not will look fine from a distance.
• Ask the learner what the last correction was and why. If they can name the reason, move 2 and move 3 are working. If they can only name what they did wrong, the reasoning is not being transferred.
• Check whether output on teaching days actually got protected. Pull the schedule for the last month and look for split pairings. Every split is a signal to the crew about what the program is worth.
• Ask the trainer, privately, whether they want to keep doing it. A trainer who is quietly resenting it will keep saying yes and slowly stop teaching, and you will read it as the program not working rather than as the trade never having been paid.

References

• U.S. Department of Labor, mentorship and supervision structure in registered apprenticeship
• Trade-standard practice for field mentoring and on-the-job instruction in skilled trades
• See related: Mentoring vs Managing: The Difference, How to Teach a Skill in the Field Without Losing the Day
• See related: The Mentor Pairing SOP, How to Let a Trainee Fail Safely and Learn From It