A BASIC Score Without the Inspection Count Behind It Is Guesswork

Two carriers. MC-4872019, DOT-2814567. MC-9103428, DOT-3657921. Both pulled on the same morning. Both sitting at the 73rd percentile in Vehicle Maintenance BASIC — above the 65% threshold that shows up as a BASIC alert and would fail most automated vetting rules I've seen. I wouldn't tender to both of them on equal terms.

The difference isn't the percentile. It's what's behind it.

MC-4872019 got to the 73rd percentile with 12 roadside inspections over 24 months. MC-9103428 got there with 87 inspections over the same period. Twelve inspections is about one every seven weeks for a carrier running moderate regional lanes. Eighty-seven inspections is a carrier that moves volume, crosses more scales, and gets looked at constantly. The BASIC score ranks both carriers in the same general neighborhood. The underlying data quality is not remotely comparable.

This is the thing nobody explains when they talk about BASIC scores.

What the percentile actually measures

FMCSA's Safety Measurement System doesn't give you a raw violation rate. It gives you a percentile rank — where a carrier sits relative to other carriers with a similar number of inspections in each BASIC category. FMCSA groups carriers into peer groups by inspection count specifically so a carrier with 10 inspections isn't unfairly ranked against one with 200. It's a reasonable statistical design.

The problem is the percentile is the number everyone stops at. They see 73rd and it reads "73 out of 100 carriers are better than this." That's roughly correct. What's hidden is that within the peer group, a carrier with 12 inspections is being ranked against carriers with up to maybe 20 or 25. Two bad inspection days in a 12-inspection sample changes the score dramatically. Two bad inspection days in an 87-inspection sample barely moves the needle.

This is not a flaw in FMCSA's methodology. It's math. Small samples have wide variance. FMCSA flags carriers with zero or near-zero inspections in a BASIC category as "Insufficient Data" — that's the visible signal for truly thin data. But the zone between "Insufficient Data" and "statistically reliable" is wide, and it's invisible in the percentile display. You can sit at the 85th percentile with 9 inspections or with 140 inspections and the number looks the same.

The variance problem in practice

Here's how this plays out at the carrier level. Take a small dry van operation — six trucks, regional distribution lanes out of Memphis, maybe 200-250 driving days per truck per year. They might get inspected eight to fifteen times in a 24-month window depending on which states they run through and how much attention they draw. Their Vehicle Maintenance BASIC is built from those 8-15 inspections.

Say two of those inspections turn up brake defects. Brake violations carry significant SMS severity weights under the Part 385 scoring methodology — brakes are the leading equipment defect in fatal truck crashes, and the scoring reflects that. Two brake-violation inspections out of 12 total puts you at a violation rate of about 17%. For a peer group of similarly-inspected carriers, 17% might land you in the 70th or 80th percentile territory. You're in alert.

Now change the scenario: same carrier, same two bad inspection days, but 87 inspections total. Two out of 87 is a 2.3% brake violation rate. Same violations. Same equipment problems. Different percentile — probably somewhere well below the alert threshold.

The two-violation count didn't change. The carrier's actual maintenance culture may not have changed. What changed is the denominator, and the denominator is mostly a function of how much that carrier runs, not how careful they are.

Why you have to look at the underlying count

Every BASIC score has a number of inspections behind it. You can see it. In the FMCSA SMS interface and on a SAFER company snapshot, the inspection count for each BASIC category is available if you look past the percentile number. Most vetting tools and most brokers stop at the alert flag. The count is one extra click — sometimes one extra column — and almost nobody goes there.

I've made it a personal rule: if a carrier is small enough that I'd be surprised to find 40+ inspections in a BASIC category, I pull the count before I make a call on the score. For Vehicle Maintenance in particular. Brake and tire defects are the thing — 49 CFR § 396.3 requires carriers to maintain vehicles in safe operating condition, and brake and lighting systems get scrutinized every time a carrier hits a weigh station. The violation data is there, but the signal-to-noise ratio at low inspection counts is rough.

What I'm looking for:

A carrier with 10-15 inspections and a BASIC alert might have had two bad weeks. It might be a chronic problem. I can't tell from the percentile alone. I need to look at the underlying violations, when they happened, and whether they cluster — all in the same month, or spread across the 24-month window. A cluster suggests a bad period that may be past. Distribution across the window is a harder conversation.

A carrier with 60-plus inspections and a BASIC alert has earned it the hard way. That signal I trust.

What the regulatory trigger actually says

49 CFR Part 385 governs safety fitness determinations and FMCSA investigation criteria. Under Part 385, carriers above certain SMS thresholds in specific BASIC categories trigger warning letters, investigations, and formal safety fitness determinations. The investigation process doesn't begin and end at 65%. FMCSA has its own prioritization logic, and small carriers with thin inspection data get weighted differently in the investigation queue.

What matters for brokers is this: Part 385 and the SMS methodology are designed to help FMCSA prioritize its own enforcement resources. They weren't designed as a binary "load/don't load" switch for brokers. When FMCSA itself uses inspection count as part of its prioritization, brokers who treat a 12-inspection BASIC alert identically to an 87-inspection alert are applying the tool incorrectly.

That distinction matters in discovery. Post-Montgomery v. Caribe Transport II (decided May 14, 2026), state-law negligent selection claims against brokers are no longer preempted. The question of what constitutes reasonable due diligence is now open to state court interpretation. A plaintiff's lawyer who subpoenas your carrier file and finds "BASIC alert → rejected" or "BASIC alert → approved, documented threshold exception" will next ask: "Did you look at how many inspections were behind that score?" If your answer is no, you've got some explaining to do.

If your answer is "yes, 12 inspections, here's the underlying violation breakdown, here's why we called dispatch and confirmed maintenance status before tender," you've added a layer of judgment that an automated rule can't provide.

The carrier I got wrong

MC-4872019 is fictional, but the pattern I'm describing isn't. I've tendered to small carriers with BASIC alerts that turned out to be noise — two bad inspections in an eight-month stretch, maintenance culture that was actually tight, and the inspection count told me I should dig instead of auto-reject. I've also held a tender on a carrier with a clean BASIC score and only 9 total inspections, because "clean" with 9 inspections is not the same as "clean" with 80.

The one I regret was the other direction: a carrier running about 12 trucks, strong broker references, BASIC scores all below alert threshold. What I didn't notice was that the Vehicle Maintenance score was based on 8 inspections. Low denominator, no violations. I took "no alerts" at face value. Later found out — not on my load, fortunately — that they'd had a significant maintenance event that never showed up in the SMS data because it predated their most recent inspections cycling out of the 24-month window.

That's a different problem (BASIC data lag is its own conversation), but the lesson is the same: the percentile tells you where a carrier ranks. It doesn't tell you how confident to be in that ranking. Inspection count is the confidence interval.

What this looks like on a real tender

Scenario: $450,000 reefer load — produce, temperature-sensitive, direct lane from Nogales to Chicago. I'm pulling a carrier, MC-6517429, DOT-4128863. Five trucks. Twelve months old on authority. Vehicle Maintenance BASIC at 68th percentile — technically in alert territory.

I check the inspection count before I pick up the phone: 9 inspections backing that 68th percentile score. I look at the underlying violations: one brake inspection, one lighting defect, both from the same week in January. Everything since has been clean.

I call dispatch. I ask about their PM schedule. I ask whether they're running these trucks on any other lanes — are they getting inspected at AZ and IL weigh stations regularly? I'm listening for whether they know what's in their safety file. If the safety director can tell me their last PM cycle date off the top of their head, that's a carrier that thinks about this. If they have to put me on hold to find out when their trucks were last serviced, I'm making a different call.

I document what I found, what I asked, and what they told me. I note specifically that the 68th percentile Vehicle Maintenance score is built from 9 inspections and that the violations clustered in one inspection period.

That record, if it ever goes into discovery, shows judgment. Not a checkbox. Judgment.

How I document this

My carrier file entry for a small-carrier BASIC review includes:

The BASIC score and percentile for each flagged category (screenshot or log date). The inspection count behind each flagged BASIC — not just the score, the count. A plain-English note on whether violations clustered or distributed, and over what timeframe. Any T-call notes if I called dispatch to verify maintenance posture. My written conclusion on why the risk is or isn't acceptable given the sample size.

That last piece is the one that takes 45 seconds to write and the one that matters most if the file ever gets opened. "Carrier at 68th percentile VM; 9 inspections, violations cluster to one period, confirmed PM schedule with dispatch, proceeding with single load" is a sentence. It's also a defensible record.

If you're auto-approving carriers with thin BASIC data because the score happened to be under 65%, you're not vetting — you're hoping the noise broke your way. Sometimes it does. Post-Montgomery, you'd rather be able to explain your process than hope.

---

— Mason Lavallet

Founder, DOTScreener.com