How to Pick Which Tech Should Be Teaching
Why this matters
The teacher assignment is made in about four seconds, usually by seniority, usually to the best tech in the shop, and it is the decision that determines whether the whole training program works. Everything downstream - the curriculum, the sign-offs, the tracker - runs through that one person's ability to make someone else competent.
Picking wrong is expensive twice. The learner develops slowly or picks up bad habits, and your most productive tech spends months at reduced output producing that result. Shops rarely connect the two, because the assignment felt obvious at the time and nobody revisits an obvious decision.
Step 1: Stop defaulting to your best tech
Skill at the work and skill at transferring the work are different capabilities, and at high levels of the first they often work against each other.
The reason is automaticity. A tech who has run a procedure ten thousand times no longer performs it as a sequence of decisions, they perform it as one motion. Ask them why they checked that point first and the honest answer is "I just do." That is a sign of mastery and a serious problem for teaching, because the decision points the learner needs are the exact things that have become invisible to the expert.
The second problem is tolerance. The best tech is usually the fastest, and watching a task take four times as long is genuinely painful for them. They take the tool back. Not out of malice, out of reflex.
None of that means seniors cannot teach. It means seniority is not evidence that they can, and it should not be treated as the qualification.
Step 2: Score the four traits that actually predict teaching
Rate each candidate 0 to 3 on these four. Twelve points possible. Do it in writing, because the point is to compare candidates rather than to confirm the choice you already made.
Narration. Can they explain what they are thinking while their hands are working? This is the single strongest predictor, because diagnostic reasoning is invisible unless it is spoken. A tech who cannot narrate can demonstrate procedures and cannot teach diagnosis.
Patience under time pressure. Not patience generally. Anyone is patient at nine in the morning on a light day. The question is what they do at the fourth call when they are behind and the learner is slow. This is where teachers are actually made or unmade.
Tolerance for a slower correct answer. Will they let the learner get there their own way if the way is safe and correct, or do they need it done their way? A teacher who cannot tolerate a different valid route produces techs who can only work under that teacher.
Willingness to be watched. A teacher is exposed all day. Their shortcuts get seen, their occasional wrong first guess gets seen, and their handling of an awkward customer gets seen. A tech who needs to appear infallible will hide their mistakes from the learner, and hiding mistakes removes the most instructive material available.
Skill in the work is a prerequisite, not a scored trait. Everyone in the candidate pool must already be signed off at teaching level on the rows in question. That is a gate you pass or fail, not a score you can be strong on to offset a 1 in patience.
Step 3: Run a two-call audition before committing
Do not decide from the score sheet alone. Put each serious candidate on two calls with the learner and observe, ideally on a moderately busy day rather than a calm one.
Watch for four specific behaviors, all of which are visible within one call:
- Do they explain before doing, or narrate while doing, or neither?
- Do they hand over the tool, and does it come back before the learner is finished?
- When the learner makes a non-critical error, what happens? Silence and a note for later is the right answer. An immediate takeover is a red flag. A correction in front of the customer is a disqualifier.
- Do they ask the learner questions, or only answer them? Teachers ask. The tech who only answers is running a help desk.
Two calls is enough. This is not a long evaluation, and the behaviors are not subtle.
Step 4: Match the teacher to the stage, not to the person
Different stages want different teachers, and the near-peer effect is real and underused.
| Learner stage | Best teacher profile | Why |
|---|---|---|
| First weeks, basic tasks and safety | A mid-level tech who learned this recently | Remembers being confused. Has not yet automated the steps. |
| Building routine competence | A steady journeyman with high patience | Consistency matters more than brilliance here. |
| Diagnostic reasoning | Whoever narrates best, regardless of rank | The reasoning is the product, and it must be spoken. |
| The ugly, non-standard cases | Your genuine expert, in short bursts | This is where deep experience is irreplaceable. |
The pattern is that your most expensive teacher should be spent last and narrowly, on the cases only they can teach, rather than first and broadly on fundamentals a mid-level tech can cover better and cheaper.
The mid-level tech being better at fundamentals is not a compromise. Someone who was confused by the same thing eighteen months ago can name the confusion. Your twenty-year tech genuinely cannot remember what it was like not to know.
Step 5: Check what the assignment costs
Before confirming, check three numbers:
Their utilization. A tech booked at or near capacity cannot absorb the teaching premium, which commonly runs around 40 to 50% longer on taught routine calls. Assigning them anyway produces a teacher who teaches for two weeks and then stops, and nobody records that it stopped.
Their unique rows. If this tech is the only one who can perform three critical tasks, every hour they spend teaching fundamentals is an hour those three tasks cannot be covered. That is a poor trade for teaching material somebody else could teach.
Their own development. Teaching is genuinely good for a tech's own depth, because explaining forces the reasoning back to the surface. If a candidate is a strong tech who has plateaued, the teaching assignment is a real development move for them and not just a cost.
Step 6: Make it a role, with a load limit and a return
An informal teaching assignment gets dropped the first busy week and nobody notices for a month. Make it explicit on three points:
- A name. Lead tech, mentor, trainer, whatever fits your shop. A named role can be held accountable and can be declined.
- A load limit. One learner at a time, and a stated number of teaching calls per week, commonly two to three. Without a cap the assignment silently expands until it collapses.
- A return. Pay, scope, title, schedule preference, or a clear path to a role that has one of those. A teaching assignment with no return is a tax on your most helpful people, and the fair-minded ones will do it anyway for a while and then quietly stop.
The return also fixes a problem people avoid naming: a tech who is the only person able to do something has a rational interest in staying that way. Paying for teaching converts that interest, so that spreading the knowledge is what advances them rather than what erodes their position.
Step 7: Handle the tech who should not be teaching
Some strong techs are poor teachers and will not become good ones on a timeline that helps you. Do not force it and do not frame it as a deficiency.
Three productive alternatives, all of which capture their knowledge without putting them in a teaching role:
- Have them write the procedure for the tasks only they run. Not a lesson, a procedure. Many people who cannot teach can document well.
- Use them as an assessor, not a teacher. Observing silently and judging a performance is a different job than instructing, and impatient experts are frequently excellent at it.
- Use them as a phone lifeline for a learner who is out on their own. Short, specific, low-patience-demand interactions are exactly what they are good at.
A worked example: scoring three candidates in a six-person shop
Three candidates, all already signed off at teaching level on the rows in question.
| Candidate | Narration | Patience | Tolerance | Willing to be watched | Total |
|---|---|---|---|---|---|
| A, twenty years, most skilled | 1 | 1 | 1 | 2 | 5 of 12 |
| B, six years, mid-level | 3 | 3 | 2 | 3 | 11 of 12 |
| C, twelve years, steady | 2 | 2 | 3 | 1 | 8 of 12 |
The default assignment would have gone to A on seniority. A scores 5 of 12, and the two 1s are on narration and patience, which are the two traits that matter most.
Then the coverage question, which is where the plan gets built. A is signed off at teaching level on 20 of the shop's 22 rows. B is at 14 of 22. So B can carry 14 rows, and there are 6 rows that only A can teach.
The plan that falls out: B is the primary teacher and carries the learner day to day. A takes a narrow assignment covering only those 6 rows, delivered as a handful of calls rather than as an ongoing pairing. C becomes the assessor for sign-offs, which suits a high tolerance score and a low willingness-to-be-watched score, since assessing is observing rather than performing.
Check it against utilization. A is booked at about 92% of available field hours, B at about 78%. B has roughly 14 percentage points of slack and A has effectively none. Assigning primary teaching to A would have required displacing billable work from the shop's most productive person to do the job they scored 5 of 12 at. Assigning it to B costs less capacity and gets a better result.
That is the whole argument, and it took about twenty minutes of scoring to reach a conclusion the four-second version got backwards.
What changes the answer
- A two or three person shop. There is no candidate pool. The owner teaches, and the practical move is to work on your own narration habit, since that is the trait most likely to be missing and the easiest to build deliberately.
- A hazard-heavy specialty. Safety-critical instruction goes to the most experienced person regardless of teaching score, with the near-peer handling everything around it. Do not optimize for teaching quality on the tasks where the failure mode is an injury.
- A learner who is not progressing. Change the teacher before concluding the learner is the problem. A pairing that is not working is often a mismatch, and testing that costs two weeks.
- A teacher whose own quality slips. Teaching load can degrade a tech's own output if the cap is not enforced. If their callback rate rises after taking the role, cut the load rather than the role.
How to verify you picked right
Ask the learner what they learned this week, and listen for reasoning. If they can only recite steps, the teacher is demonstrating and not narrating, whatever their score said.
Check whether the tool comes back. Ride along on one call and watch for the takeover. A teacher who takes the tool back before the learner finishes, on non-critical work, has not actually accepted the role.
Watch the teacher's own numbers. Their call count will drop and should, by roughly the teaching premium times the number of taught calls. If it drops much further, something else is wrong. If it does not drop at all, no teaching is happening.
References
- See related: Mentoring vs Managing: The Difference, Being a Mentor: The Give-Back
- See related: The Apprentice-to-Journeyman Path as a System, How to Teach a Skill in the Field Without Losing the Day
- OSHA requirements that employees performing hazardous work be trained by and work under qualified personnel
- Trade-standard practice for mentor selection in registered apprenticeship programs