How the score works
The DT Score is DollarTraq's measure of how safe a motor carrier is to book, from 0 (do not book) to 100. It is calculated from public FMCSA data, and it comes with the reasons behind it, so a person can check the call.
What is checked#
The model runs about fifty checks, grouped into seven areas. Each response says how many ran, found a problem, passed, or could not run (checks), and lists every problem found (reasons).
| Area | What it looks at |
|---|---|
| Authority & Compliance | Operating authority status, revocations and reinstatements, suspensions, out-of-service orders, pending applications |
| Insurance & Financial | Liability (BIPD) and cargo insurance on file, cancellations and gaps |
| Safety & Roadside | FMCSA safety measures against intervention thresholds, out-of-service rates |
| Crash History | Crashes, fatalities, injuries and tow-aways in the last 24 months |
| Inspection Quality | How often roadside inspections find violations |
| Identity & Fraud | Phone numbers, emails, addresses and vehicles shared with other carriers |
| Operations & Experience | Age of the authority, how current the MCS-150 filing is, fleet size against what is seen on the road |
From findings to a score#
Every carrier starts at 100. Each check that finds a problem adds risk according to how serious it is, from a minor finding to one that needs a person's judgment. The total risk is mapped onto the 0–100 scale, so the score and the decision always agree.
Two things can hold a score down further:
- A knockout caps the score at 18 and makes the decision
rejected, whatever else the carrier gets right. Knockouts are findings that mean the carrier cannot legally or safely haul, such as inactive operating authority or no liability insurance on file. They are listed inknockout.reasons. - Thin data caps the score. Missing data never counts as clean and never counts as bad; a check that cannot run is reported in
checks.not_checkedinstead. When too little is known, the score cannot rise high, anddata_confidenceshows how much was known.
Grades#
| Grade | Score |
|---|---|
| A | 90–100 |
| B | 80–89 |
| C | 70–79 |
| D | 60–69 |
| F | 0–59 |
Decisions#
The decision is the field most integrations act on. It follows the findings, not the number alone.
| Decision | When | Suggested action |
|---|---|---|
| approved | Score 80 or more, and no finding that needs a person | Book |
| review | Score 60–79, or any finding that needs a person's judgment, whatever the score | Have someone read reasons before booking |
| high risk | Score below 60 | Avoid, or book only with extra safeguards |
| rejected | A knockout fired | Do not book |
Because a serious finding forces review, a carrier can be in review with a score below 60. Branch on decision, not on score ranges you work out yourself.
Manual review and data confidence#
needs_manual_review is true when a person should look before the carrier is booked, even if the decision is not review. The most common cause is thin data: data_confidence (0 to 1) is the share of the core inputs that were available, and below 0.65 the score is always flagged. New carriers often land here because they have little history yet.
How fresh a score is#
The FMCSA data behind the score is refreshed daily. A score is recalculated whenever it is requested, and reused for up to 15 minutes for the same carrier; calculated_at says when it was calculated. Each response also names the model_version that produced it. See Versioning for how model changes are announced.
What the score is not#
The DT Score is a decision aid built on public records. It does not replace your own carrier vetting, contracts or insurance checks, and it is not a guarantee of how a carrier will perform. Records can lag reality; when a decision matters, read the reasons and confirm the facts behind them.