How to Train When You Are Too Busy to Train

Why this matters

Every shop that says it has no time to train is telling the truth about its calendar and lying to itself about its math. The busy week is exactly the week the training deficit shows up: the senior tech gets pulled off a paid call to walk a second-year through something over the phone, a job runs long because the only person who can do it is already on another roof, and a callback lands because someone did work they were never actually taught. Training does not compete with billable hours. It competes with the hours you are already losing to the gap, you just do not have a line for those hours anywhere.

The fix is not a training day. A training day is the first thing that gets cancelled, and it gets cancelled in July when you need it most. The fix is to stop treating training as a thing you schedule and start treating it as a property of jobs you were already going to run.

Step 1: Set the budget as a percentage of hours, not a calendar block

Pick a number between 3% and 8% of your total billable field hours and call it the training budget. Under 3% and nothing measurable happens. Over 8% in a shop with no bench and you will feel it in the schedule inside a month.

The reason it has to be a percentage and not a block is that a percentage scales down automatically when you get slammed and scales up when you do not, without anyone having to make a decision. A four-hour Friday block gets cancelled the first week you are behind, and once cancelled it never comes back. A 5% budget in a bad week is just a smaller number of paired hours, which is survivable, so it survives.

Write the number down and tell the crew what it is. A tech who knows the shop spends 5% of its field hours on teaching reads a paired job as policy. A tech who does not know reads the same paired job as "the boss thinks I am slow."

Step 2: Pick the jobs that teach instead of creating jobs that teach

The single highest-leverage move is job selection. You already run enough variety in a month to cover most of a competency list. What you do not do is choose which body goes to which job on the basis of what it teaches.

Sort your recurring job types into three buckets:

• Teaching jobs. Repeatable, moderate complexity, low customer-pressure, forgiving of a slower pace. Maintenance visits, straightforward replacements, planned installs with a wide arrival window. These carry a trainee at almost no risk.
• Watching jobs. High complexity or high stakes, where the trainee gets real value from observing but must not touch. Anything with live hazards, anything on a difficult customer, anything on a tight completion deadline.
• Do-not-teach jobs. Emergency calls, anything already running late, anything where the customer is angry before you arrive. Teaching on these is how you turn one problem into two.

Then have whoever dispatches make one extra decision per day: which of tomorrow's teaching jobs gets a second body on it. That is the whole intervention. It costs a dispatcher about two minutes and it is the difference between a shop that trains and a shop that talks about training.

Step 3: Load the pairing where the drag is cheapest

Pairing is not free. A senior tech running a job with a trainee alongside runs it slower, because narrating what you are doing and watching someone else's hands both cost attention. In practice, expect a paired job to run roughly 20% to 30% longer than the same senior would take solo, higher at the start of a trainee's exposure to a job type and dropping fast once they have seen it twice.

That drag is the actual price of training, and it is much smaller than the price of the training day you keep cancelling. But it is real, so put it where it hurts least:

• Front-load the week. Monday and Tuesday absorb an overrun. Friday afternoon does not, because the overrun becomes unpaid overtime or a bumped job.
• Pair on the first job of the day, not the last. The overrun eats into slack, not into the customer at the end of the route.
• Never pair on the day's tightest route. If the day already has four stops across a wide area, the drag compounds through every remaining stop.

Step 4: Use the hours you already pay for and do not bill

Every shop has non-billable paid time it has never audited: drive time between stops, the wait for a supplier counter, the twenty minutes at the end of a day spent restocking a truck, the delay waiting for a customer to get home. In a shop with meaningful drive distances this is often the largest untapped training block available, and you are already paying for it.

Drive time is the best of these, because it is protected, private, and long enough for real teaching. It also has a natural curriculum attached: you have just left a job, so debrief that job. What did we find, what did we rule out, and why. What did I decide that you would have decided differently.

The rule that makes this work is that the senior asks first and answers second. "What do you think that noise was" produces learning. "That noise was the blower wheel" produces a fact that will be gone by Thursday.

Step 5: Give each month one named target

A busy shop cannot run a broad curriculum. It can run one target. Pick a single competency per month per trainee, name it in a sentence a tech would actually say, and let job selection serve it.

"This month you own the full startup sequence, including the readings and the paperwork" is a target. "Improve diagnostics" is not, because nobody can tell on the last day of the month whether it happened.

One target per month is roughly 8 to 12 exposures at two paired jobs a week, which is enough repetition for most single competencies to stick and short enough that the trainee can feel the finish line. Two targets at once halves the repetitions on each and usually delivers neither.

Step 6: Make the trainee do the capture

The senior tech's time is the scarce resource, so do not spend it on documentation. The trainee writes the notes: what the job was, what the sequence was, what surprised them, what they would do differently. The senior spends two minutes reading it, not twenty minutes writing it.

This is not an administrative preference. Writing the sequence down forces the trainee to notice the steps they cannot actually account for, which is exactly the list you want for next week's pairing. It also builds the beginnings of a shop reference that outlives whoever wrote it.

A worked example: the three-tech shop with no slow week

Take a shop with three field techs. Each is paid 40 hours a week and puts about 32 of those on billable work, so the shop runs 96 billable field hours a week. The trainee is the second-year; the teacher is the senior.

Setting the budget. At 5%, the training budget is 4.8 hours a week of field time, call it 5. That is not 5 hours of classroom. It is the amount of senior capacity the shop is willing to lose to pairing.

Spending it. Two paired jobs a week, each a job the senior would run solo in 3 hours. Paired, they run about 25% long, so 3.75 hours each. The drag is 0.75 hours per job, 1.5 hours a week of lost senior capacity. Against 96 billable hours that is about 1.6% of shop capacity, well inside the 5% budget, and it leaves headroom for the weeks that go sideways.

The trainee's side of the ledger. The trainee is on the clock for 7.5 hours across those two jobs regardless. Before the change, those hours were spent fetching and holding. After, they are spent on a named competency.

The payback. Say it takes 8 paired repetitions before the trainee runs that job type solo. At two a week that is 4 weeks, and the total investment is 4 weeks times 1.5 hours, so 6 hours of senior capacity.

Now the trainee takes that job type solo. Early on they will run slower than the senior, maybe 30% slower, so 3.9 hours against the senior's 3. That slower time is not a loss, because it is the trainee's hours, not the senior's. In the first week of solo work the trainee takes both of those jobs, which frees 6 hours of the senior's week outright.

The 6 hours you spent come back in the first solo week, and every week after that is profit. Run the same loop on a second job type the following month and the effect compounds, because the senior now has 6 free hours a week to spend teaching the next thing.

What the same shop looks like without this. The senior stays the only person who can run that job type. When the senior takes a week off, those jobs get pushed or subbed. When the senior gets a better offer, the shop loses a service line. The 6 hours were never saved, they were deferred at interest.

What changes the answer

A two-person shop. With one senior and one trainee there is no third body to absorb the drag, so the 5% budget realistically drops to 3% and the pairing has to land on the least time-sensitive job of the week. It still works, it just moves slower, and you should expect a competency to take 6 to 8 weeks rather than 4.

A trainee who is not new to the trade. Someone with real experience from another shop needs verification more than instruction. Pair them for two jobs, not eight, then move to a spot-check pattern: they run it solo, the senior reviews the documented result. Running a fully experienced hire through an eight-rep sequence reads as distrust and you will lose them.

Genuine seasonal collapse. If your slow season is real, some of this belongs there instead. Use the slow weeks for the things that genuinely need a block of uninterrupted time, and keep the job-embedded loop running year round for everything else.

A safety-critical competency. Anything involving gas, stored energy, live electrical work, or working at height does not go on the 8-rep casual loop. Those get formal instruction, a documented sign-off, and direct supervision until signed off, regardless of what it does to the week. The budget bends for capacity, not for hazard.

How to verify you got this right

Three checks, all cheap:

• Count paired jobs, not training hours. Pull the last four weeks and count how many jobs had a second body assigned for teaching. If the number is zero in a busy month, the policy is not real. If it matches your budget in a bad month too, it is.
• Watch the dispatch grid for single points of failure. Take any job type and ask who can run it alone. If the answer is one name for more than two job types, your training is not reaching the bottleneck, it is reaching whatever was convenient.
• Track time-to-solo per competency. If the second competency does not take fewer weeks than the first, the trainee is collecting exposures without building a method, and the drive-time debrief is probably being skipped.

The failure mode to watch for is the paired job that is really just a two-body job. If the trainee spends the visit fetching parts and holding a light, the shop paid the drag and got a helper, not a trainee. The tell is the debrief: if the trainee cannot describe why the senior made a decision, they were labor, not a student.

References

• See related: The Ride-Along Sequence That Actually Builds Competence, Build a Skills Matrix: Who Can Do What
• See related: The Off-Season Systems and Training Push, Building a Training Culture Not Just Onboarding
• OSHA general industry training requirements for hazard-specific tasks
• Trade-standard practice for field-service apprenticeship supervision