The last guy I hired had eleven years on roofs and could read a slope better than two of my longer-tenured people. His first solo inspection was still unusable. Not wrong, unusable: no wide shots, no south elevation at all, and a two-minute exchange in the driveway where a homeowner asked about her deductible and he answered. When you train new inspector hires, the roof knowledge is usually the part you did not have to buy. The part you have to build is the record.
So I stopped running week one like an apprenticeship and started running it like a production standard. Five days, three phases, one measurable target: can this person produce an inspection record that looks like everybody else's.
Day one and two: he watches me, and he narrates
Ride-alongs are standard and mostly wasted, because the new guy stands there absorbing atmosphere.
I changed one thing. On day one he watches. On day two he does the talking while I do the climbing. I run the route, he narrates the walk out loud to the homeowner or to the camera. It is uncomfortable for about twenty minutes and then it is the fastest teaching tool I have found, because the gaps show up in his mouth instead of in a report three weeks later. He will say "there's damage here" and I will say what kind, and he will say "hail" and I will say how do you know, and by the fourth house he is saying "bruising, about the size of a nickel, granules knocked loose in the middle" without being prompted.
Two inspections a day is plenty. Four is theater.
Day three: he runs it, I stand on the ground
This is the one people skip and it is the whole week.
He takes the ladder, the card, and the phone. I stay on the ground and do not coach. Not one word, even when he goes counterclockwise, even when he forgets the wide shot on the back slope. I write it down and we go through it in the truck after.
The first solo walk is a diagnostic, not a performance. If I coach through it, I learn nothing about what he actually retained, and he learns that somebody will catch it, which is exactly the habit I am trying not to install.
Three things go wrong on almost every day three. The route drifts under time pressure. The wide shots vanish because close-ups feel more useful. And the narration goes silent for long stretches while he concentrates, which turns a live session into a shaky home movie for whoever is watching.
All three are fixable in one conversation because they are habits, not judgment.
Day four: train new inspector language, including what he does not say
Half a day, in the truck and on one live roof, on nothing but language.
The first list is the vocabulary: bruising, granule loss, backed-out nail, lifted ridge cap, cracked boot, soft decking. Say it the same way every time, so a homeowner who gets three inspectors from my company over five years hears one company instead of three guys.
The second list is what he does not say, and I spend more time on this than any other single item when I train new inspector hires. He documents condition. He does not tell a homeowner whether something is covered, what a claim might pay, whether to file, what their policy means, or what an adjuster is likely to do. That is not modesty and it is not a disclaimer I read to him. I am a contractor and not a licensed public adjuster, and in most states doing that work unlicensed is a crime. Coverage is between the homeowner and their insurer, full stop, and we do not name insurers or speculate about them on a roof.
Then we drill the replacement sentence until it comes out without thinking: "I document what I can see and it all goes in the record. Coverage questions go to your insurer, and this footage is yours to send them."
Homeowners ask. They ask on almost every job, usually while you are coming down the ladder and your hands are full, which is the worst possible moment to be composing an answer. That is why it gets drilled instead of mentioned.
Day five: two solos and a review
He runs two by himself. I review both that evening, against five things and nothing else.
Did the route run in order. Do the wide shots exist for every slope. Is every penetration captured, including the ones that were fine. Was the narration continuous enough that somebody watching knows where they are. Did anything come out of his mouth that belongs to an insurer.
That is the week-one bar. Notice what is not on it: whether he found everything. He will miss things for six months, same as I did, and that gets fixed by reps and by the review loop rather than by day five. What cannot wait six months is the shape of the record, because that goes to a customer immediately.
Being able to review both of those inspections from my kitchen table instead of driving back out to the property is the only reason this schedule is possible at all. Before that, training a second inspector meant I lost most of a week of my own production, which is precisely why I put it off for two years. The shops on the use cases page that run this tightest tend to be the ones where the owner reviews footage at night instead of shadowing all day.
The review habit itself is worth building deliberately, and it is a different discipline from training. I covered it in how to spot-check inspections without becoming the bottleneck, and the whole system this fits into is in running inspections across multiple crews.
When to stop
Some people do not get there, and the tell is not competence.
The guy who misses a soft spot is teachable. The guy who cannot stop improvising the route, or who keeps answering questions that are not his to answer after being told twice, is telling you something about how he will behave when nobody is reviewing. Two weeks is enough to know. I have kept people longer than that out of stubbornness and it cost me more than the hire did.
If you are about to train new inspector number two or three, decide the week-one bar before they start rather than after. I am glad to walk through how other shops phase a rollout, and the pricing is per inspection so a trainee running two walks a day for a week does not blow up your bill.
InspectStream records what an inspector observed. It is not a public adjusting service and does not prepare, negotiate, or advise on insurance claims.