Every commercial insurance renewal starts the same way. The underwriter pulls your loss history, plugs your vehicle count into a formula, and hands back a number that has very little to do with how your people actually drive. Your team gets priced like an average fleet. If you show up with your own data, already sorted the way an underwriter reads it, you get to argue from evidence instead of arguing against a number somebody else built.
I came at this sideways. A few times a year I plug a BlueDriver OBD-II reader into my own car and go through the log (hard brakes, mostly, and how heavy my right foot gets on the highway), mostly because I like knowing what the sensors think of me. At some point I noticed that the things I was checking for fun are close to what commercial insurers look at when they price a fleet. Fleet managers usually have that data already. What they rarely have is a version of it an underwriter can use.
Safety scores, and the trend behind them
A safety score rolls individual driving events into one number. Most telematics platforms build it from hard braking, rapid acceleration, speeding against posted limits and sharp cornering, though each vendor weights those differently. The number on its own says less than its direction. A team whose score has climbed over twelve months tells an underwriter that somebody is paying attention, and a high score that never moves could just be a quiet year.
Loss history only records what went wrong. A score trend also covers the ordinary days in between, which the underwriter has no other way to see.
The format question is where I’m still unsure (and I suspect every underwriter answers it differently). Some want CSV exports broken down per driver so they can sort it themselves, while others would rather have a short PDF with a trend chart. Whichever they prefer, they end up asking how you define a hard brake and how much driving the data actually covers. If your report answers those before anyone asks, it gets read differently from a printout with no explanation of where the numbers came from.
Bring driver-level numbers, not only the fleet average.
An average hides the spread between your best and worst drivers, and underwriters tend to ask about the spread early. If you only show the mean, expect them to assume the spread is bad. When a couple of drivers really are pulling the number down, I would rather say so up front and explain what you’re doing about it (coaching, say, or a change of route) than have the underwriter find it in the CSV.
Mileage: estimates cost you money
Commercial auto premiums lean partly on estimated annual mileage, and most operators submit a rough figure at the start of the policy, usually rounded up or copied from last year. That guess then sits inside your premium for the whole term. If you drove less than you estimated, you paid for miles you never covered. Drive more than that and the audit catches it and you get a bill you weren’t expecting.
GPS trip records give you an actual distance for every working day. Nobody can check a guess, so it tends to get a safety margin added on top, while a measured total that comes with a note on how it was collected is usually taken much closer to face value.
This is the part my hobby actually helps with. GPS mileage is the sum of the straight-line distances between location pings, so it depends on how often the device records a point. A system that logs every thirty seconds cuts corners, and over a year of driving those missed curves add up to a lower total than the odometer shows. Phone-based trackers that update every few seconds while moving stay much closer to the odometer. If the underwriter asks how you measured mileage, you want to be able to name the update interval without going to look it up. And if you can’t find it in your platform’s documentation, ask the vendor before the meeting, not during it.
Raw totals are a start. Mileage broken down by when and where it was driven is more persuasive. Say most of your team’s driving happens in daylight on the same regular routes, with a small share at night or on unfamiliar highways. Showing that split gives the underwriter a concrete reason to reconsider your rate class. This one takes more work than the others, and if your routes change every week it may not show much of a pattern at all. Most platforms export trips with timestamps, so sorting them by time of day is an afternoon in a spreadsheet, and very few fleets bother.
When something goes wrong on the road
After an incident, the quality of your records shapes how the claim is handled and how it sits in your loss history afterwards. Timestamped location history gives a record that doesn’t depend on anyone’s memory of a stressful afternoon. That’s useful for the driver as much as for you. If the person was on the assigned route and stopped where they said they stopped, the data backs them up.
The adjuster’s questions tend to be plain ones. Where was the person, and when? Were they on the planned route or somewhere else? Some hardware telematics units also record speed at each point, which answers more, though not every tool keeps it. Having those answers pulled from trip data the same week, instead of pieced together from memory later, changes how cleanly the claim gets categorized.
Replay helps here because adjusters are not GIS people. A trip played back on a map, with dots moving along the roads, is something anyone can follow without reading a spreadsheet of coordinates. Before you lean on it, check how long your tool keeps replays (some keep them for weeks, some for much longer). If the retention window is short, save or screenshot what you need right after the incident. Records you had and then lost look worse in a claim than records you never had, because someone will ask why they’re gone.
What gets an underwriter’s attention
From what I’ve read and heard from people who sit on the fleet side of renewals, hard braking events per hundred miles comes up again and again as the metric underwriters ask about first. It’s commonly cited as tracking collision risk better than speeding alone, though I’d ask your own broker how their carriers weigh it rather than take that from me.
Mileage only gives you something to negotiate with when it beats your original estimate. If it matches what you reported, the premium already reflects it. Driver turnover and tenure come into it too, even though they’re HR numbers rather than tracking data. A team where people stay is usually read as a steadier risk.
What they skip
Fuel use and idle time don’t change your premium. They matter for running the business, and I’d keep tracking them, but they say nothing about collision or liability risk. A thick binder full of operational charts mostly makes the underwriter dig for the pages they wanted. Keep the insurance package short. If you’re not sure whether a chart belongs, ask yourself whether it would change the underwriter’s guess about how likely your team is to have a crash next year. When the answer is no, it goes in the appendix or stays home (I know, it’s hard to leave out a chart you spent a weekend on).
Sending it before renewal
Most carriers seem happy with a summary PDF plus the raw CSV attached. The PDF shows the trend, and the CSV lets their analysts check your numbers themselves.
Don’t wait for renewal. Send a short report every quarter, so by the time the renewal conversation starts, your underwriter has already watched months of data come in. At that point the conversation is about whether the rate matches risk they’ve already watched being measured.
If your team already shares location from their phones, you may have more of this than you think. The trip replays from those drives show where each person went and when, without adding hardware to anything. They won’t give you a braking score (phone location data doesn’t record that), and replay links often expire after a few weeks, so save the ones tied to an incident and pair them with whatever safety data you keep longer. Before your next renewal, export last quarter’s mileage from whichever system logs it and write a paragraph explaining how it was measured. If your insurer asks for more, you’ll know exactly what’s missing.