How our numbers were measured

Every statistic Carrier Archive publishes comes from one measured, dated test. Measured 2026-08-08; this page updated 2026-10-26.

The measured result

22.4% vs 11.0%

failure rate of flagged carriers vs all carriers — measured on 166,237 held-out carriers

Test population: 166,237 carriers with active for-hire authority (September 2025 records), none of them used during development.
Our record-pattern indicators flagged 4,280 carriers (2.6%).
Over the following 11 months, 22.4% of flagged carriers lost their operating authority involuntarily (95% confidence interval: 21.2%–23.7%), versus 11.0% of all carriers in the population.
That is at least 1.6× the baseline rate by point estimate. The result was statistically significant (p = 0.0001).

The rules were frozen in writing before the test, and the test ran once, on carriers our development never touched.

What these numbers mean — and don't

These are observed associations measured on historical data, not predictions about any individual carrier. An indicator on a carrier's file is a documented record pattern with a measured historical rate — it is not an accusation, and most flagged carriers did not fail. We publish the measured rates with their confidence interval and nothing stronger.

Measurement completed 2026-08-08 under a frozen, signed protocol. Federal files checked daily. Questions about the data itself: see the FAQ or Legal & Insurance.