Methodology

How fleetfax reads the public safety record

Every fleetfax report is built from the public federal record. This page explains, in plain terms, where the numbers come from, what the safety percentiles mean, when we do not compute one, and how to correct a number you think is wrong. It is written for the person reading a report, not for a data engineer, so it stays at the level of concepts.

Two things to say up front. fleetfax is descriptive: it reports what the record shows and puts the evidence in front of you. It never tells you to book or not book a carrier, and it computes no single composite safety score. And fleetfax is not affiliated with FMCSA. It reads public FMCSA data and is independent of the agency and the U.S. Department of Transportation.

Where the data comes from

fleetfax reads the public federal record that FMCSA publishes about every registered carrier: registration and company details, operating authority, insurance filings, roadside inspections, crashes, and safety ratings. We pull those public datasets into our own backend and refresh them daily, which is why a report loads in seconds instead of making you wait on a government site.

Freshness is ultimately bounded by the source. FMCSA publishes different datasets on different cadences, and no tool built on the public record can be fresher than the record itself. Where a figure is tied to an older filing, the report says so.

The 24-month window

The safety signals in a report are computed over a rolling 24-month window, the most recent two years of the record. This is the same window FMCSA's own Safety Measurement System (SMS) uses for its safety calculations, so a fleetfax number looks at the same recent slice of a carrier's history that the federal system looks at.

Comparing a carrier to similar carriers

A raw count is hard to read: a large fleet racks up more inspections and more crashes simply by running more trucks. So the safety picture is expressed as a percentile, meaning where a carrier sits relative to other carriers rather than an absolute count.

FMCSA's SMS handles this by sorting carriers into safety event groups, so a carrier is compared against others with a similar amount of activity on the road rather than against the entire industry. fleetfax follows the same peer-grouping idea. A fleetfax percentile answers "how does this carrier compare with similar carriers," not "how does it compare with everyone."

What "fleetfax peer estimate" means

You will see fleetfax percentiles labeled a fleetfax peer estimate. The label exists because of a legal quirk: under the FAST Act of 2015, FMCSA removed the official SMS percentiles from public view for all seven of its BASIC safety categories. Those official percentiles exist inside FMCSA, but nobody outside the agency can see them.

So fleetfax computes its own percentile from the same public inspection and crash records FMCSA scores against, and labels it plainly as an estimate. A fleetfax peer estimate is our best estimate of a number FMCSA keeps non-public. It is never presented as an official CSA, SMS, or FMCSA score, because it is not one.

When we do not estimate a percentile

fleetfax only computes a peer estimate when a carrier has enough of the underlying public record to make the comparison meaningful: enough recent inspections or crashes, and current enough information on file. This mirrors FMCSA's own published practice of requiring a minimum amount of safety activity before a carrier is measured against its peers.

When the record is too thin or too old to support an estimate, fleetfax shows the underlying facts (the counts and the records themselves) and no percentile. We would rather show you the raw record and say "look closer" than publish a number the data cannot support.

The crash rate

Crashes are shown as a rate, not a raw count, so a small carrier and a large carrier can be read on the same scale. The rate is per mile driven: the federally recordable crashes in the recent window, divided by the miles the carrier reports driving, stated per million miles. "Federally recordable" is FMCSA's own threshold for which crashes count: a fatality, an injury requiring medical attention away from the scene, or a vehicle towed from the scene.

One caveat: the mileage comes from the carrier's own federal registration filing (the MCS-150), so the rate is only as good as that self-reported number. fleetfax shows the period the mileage covers and labels the rate as historical, so it is not mistaken for today. For context, a federal benchmark of roughly 0.96 crashes per million miles (from FMCSA's Safe Driver Apprenticeship Pilot Program, 2022) is shown alongside the rate.

A report is evidence, not a verdict

fleetfax surfaces facts and flags; it does not make the decision for you. It never says a carrier is safe to book or unsafe to book, and it rolls nothing up into a single composite safety or quality score. The percentiles, the crash rate, and the risk flags are evidence you weigh. The judgment, and the responsibility for it, stays with the broker.

Found a number you think is wrong?

Because the record is public, you can always check a fleetfax figure against the source. The fastest route is the Feedback button on any report, or an email to [email protected]. A real person reads every one.

You can also compare against FMCSA's own view of the carrier's safety data at FMCSA's Safety Measurement System. If the underlying federal record itself is wrong, FMCSA's DataQs is the official channel to request a correction at the source. A correction there flows back into everything built on that record, fleetfax included.