Every contractor list is a photograph of a moving object. The question that matters — and that almost nobody selling data will quantify — is how fast the object moves. We can quantify it, because our pipeline snapshots the license status of every record we track (1,011,011 records across 39 state license tables at the latest pass) and logs a transition event whenever a status changes between observations.
Between August 7 and September 2, 2026, the observer logged 13,276 transitions: 8,514 licenses lapsed out of active status, 4,652 previously-lapsed licenses returned to active, and 110 changed status in other ways. That is real, observed motion — not an estimate.
Where the motion was
California dominated the window: 6,651 lapses and 4,196 reinstatements. That is partly scale (California's roster is our largest at roughly 238,000 records) and partly cadence — CSLB licenses renew on two-year cycles that expire throughout the year, so every month a slice of the roster hits its renewal date and a fraction of it misses.
Virginia logged 888 lapses, Minnesota 570 lapses against 310 reinstatements, Florida 405 against 146. States absent from the list mostly weren't re-observed during this particular window — transitions only become visible when we observe a record twice, so these figures are a floor on the true motion, never a ceiling.
Half the losses come back
The single most useful ratio in the data: for roughly every two licenses that lapsed in the window, one previously-lapsed license returned to active. A large share of "lapsed" contractors are not out of business — they are late on paperwork, mid-renewal, or waiting on an insurance certificate. They come back.
This is why we keep lapsed records instead of deleting them, and why our re-ingests never overwrite contact data with blanks (a database rule enforces it). A contractor who lapses in March and reinstates in May is the same business with the same phone number the whole time. Deleting the record at the lapse would mean re-buying data we already verified.
What churn means if you buy lists
Extrapolate conservatively and the arithmetic is stark: thousands of status changes per month across a million-record universe means a list bought in January is measurably wrong by June — not because anyone was sloppy, but because the underlying registries move. This is the mechanical reason our delivery pipeline re-checks stale records at purchase rather than shipping whatever was cached, and why records untouched for 180 days get re-verified before they reach a buyer's CSV.
It is also the honest argument against "lifetime" datasets sold as one-time downloads elsewhere: the download is fine; the calendar is the problem. Status data in this guide comes from our own observation log and updates continuously — the numbers above describe the stated window, and we will re-report as the log accumulates history.