Live measurement
FMCSA insurance feed status
FMCSA is currently publishing 1,774 forward-dated insurance cancellation filings, against a pre-rebuild baseline of about 1,292 per day. We have measured this number every day since 2026-06-11, which is why the 2026-08-04 break below is a record rather than a reconstruction.
By the XDate Alert team — built on public FMCSA / MOTUS records and federal regulations, reviewed against primary sources.
Last updated .
What the numbers say today
Scroll across to see every column.
| Forward-dated filings in the source | 1,774 |
|---|---|
| Pre-rebuild baseline (median of the 14 days before 2026-08-04) | 1,292 |
| Our leads the live source still confirms | 320 |
| Our leads it no longer confirms (withheld) | 83 |
| Days measured | 89 |
Observation record — discontinuity of 2026-08-04
Record v1.0, published 2026-08-09. The figures in this section are frozen: they are what we observed, and they do not change when the page regenerates. The live series above does.
We observed a discontinuity in the published MOTUS insurance data. We are not in a position to say what caused it, and this record makes no claim about intent, fault or whether the change was a defect or a deliberate correction. What follows is what we measured.
Scroll across to see every column.
| Dataset | 3uet-3z4i on data.transportation.gov |
|---|---|
Dataset rowsUpdatedAt when the change appeared | 2026-08-04T10:29:11Z |
| Sibling dataset rebuilt in the same window | 2026-08-04T10:29:32Z |
| Forward-dated count, day before | 1,300 |
| Forward-dated count, day of | 106 |
| Baseline (median, 14 days prior) | 1,292 |
A large block of filings carried a single cancellation date, 20260805, and grew over the following days: 14,134 rows on 2026-08-06, 14,325 rows on 2026-08-07, 14,441 rows on 2026-08-08. Policy effective_date values inside that block span 2017–2026.
Reconciliation of stored records against the live source
Method: draw a random sample of our stored leads that were still active with a cancellation date in the future; for each, query the source for max(cancl_effective_date) for that USDOT number; classify as matched, absent (no filings returned for the USDOT), or a different date. Two independent samples were drawn on the same day.
Scroll across to see every column.
| Sample | Matched | Absent | Different date | Observed |
|---|---|---|---|---|
| 40 | 0 | 24 | 16 | 2026-08-06 |
| 40 | 1 | 19 | 20 | 2026-08-06 |
The sampled USDOT numbers are not published here. We publish aggregate counts only. A researcher, journalist or the agency who wants the record-level sample in order to check this independently can have it — write to us and say who you are and what you intend to test.
Why this is easy to miss downstream
Nothing errored. Stored records kept carrying cancellation dates the source no longer held, and a pipeline that checks only whether its daily run succeeded would have reported healthy throughout. The failure is silent by construction: a filing whose date changes upstream does not overwrite the old row, it arrives as a different one.
Reproduce it
The forward-dated count on any given day is a single query against the public dataset — no account, no key:
https://data.transportation.gov/resource/3uet-3z4i.json
?$select=count(1)
&$where=cancl_effective_date >= 'YYYYMMDD'Our full daily series is available as CSV (aggregates only, CC BY 4.0).
Limitations
- Single observer, one reading per day, taken shortly after 18:00 UTC. A different sampling time could produce different figures around a refresh.
- We observed the published dataset, not FMCSA’s internal system. A filing missing from the export may still exist upstream.
- We do not know whether the change was a defect, a partial reload, a deliberate correction, or a change in publication semantics. We have asked; we have no answer to report.
- Whether the pre-rebuild dates or the post-rebuild dates are the correct ones is not something we can determine from the outside.
- The reconciliation samples are small (40 each) and were drawn from our own stored inventory, which is filtered — four states are excluded at ingest — so it is not a random sample of the whole dataset.
Corrections
If any figure here is wrong we will correct it on this page and note what changed and when. Nothing has been corrected since publication on 2026-08-09.
Daily readings
Scroll across to see every column.
| Date | Forward-dated filings | New to us |
|---|---|---|
| 2026-08-21 | 1,220 | 144 |
| 2026-08-22 | 1,266 | 129 |
| 2026-08-23 | 1,233 | 59 |
| 2026-08-24 | 1,223 | 63 |
| 2026-08-25 | 1,335 | 156 |
| 2026-08-26 | 1,431 | 148 |
| 2026-08-27 | 1,691 | 167 |
| 2026-08-28 | 1,696 | 135 |
| 2026-08-29 | 1,617 | 71 |
| 2026-08-30 | 1,639 | 62 |
| 2026-08-31 | 1,607 | 60 |
| 2026-09-01 | 1,806 | 269 |
| 2026-09-02 | 1,745 | 154 |
| 2026-09-03 | 1,718 | 142 |
| 2026-09-04 | 1,730 | 130 |
| 2026-09-05 | 1,715 | 110 |
| 2026-09-06 | 1,693 | 53 |
| 2026-09-07 | 1,668 | 46 |
| 2026-09-08 | 1,657 | 80 |
| 2026-09-09 | 1,731 | 149 |
| 2026-09-10 | 1,774 | 162 |
Method
Once a day we query the FMCSA/MOTUS insurance dataset on the U.S. DOT open-data portal for filings whose cancellation date is today or later, and record the count. The figure is the raw source count before any filtering of our own, so it measures the source rather than our product.
“Confirmed” means the most recent ingest saw the filing again. Because our unique key includes the cancellation date, a date that changes upstream produces a new row and leaves the old one untouched — so a filing the source has dropped is visible as one that stopped being re-confirmed. Anything unconfirmed is withheld from customers.
We publish aggregate counts only. No individual carrier records appear on this page.
Frequently asked
Frequently asked questions
What happened to the FMCSA insurance data on 4 August 2026?
FMCSA rebuilt the MOTUS insurance datasets. The number of forward-dated cancellation filings visible in the source fell from roughly 1,292 to 106 within a day, and a large block of historical filings was re-stamped with a single cancellation date. We measured the count before, during and after, because we had been recording it daily since 2026-06-11.
Is the FMCSA cancellation feed working now?
Partially. The source currently publishes 1,774 forward-dated filings against a pre-rebuild baseline of about 1,292 — at or above it. Genuinely new filings are arriving again, so the figure is rebuilding day by day, but the stock of advance notice has not returned to its former level.
Why does the forward window matter more than the total row count?
Under 49 CFR 387.313(d) an insurer must notify FMCSA at least 30 days before cancelling a motor carrier's required coverage, and that notice is public the day it is filed. The value of the data is entirely in that advance window. A dataset can hold millions of rows and still be useless for prospecting if none of them are dated in the future.
Can I build my own feed from this data instead of buying one?
Yes — the filings are public and free. Fetching them is straightforward. The hard part is noticing the day the source changes underneath you: on 4 August a pipeline built the week before would have carried on emailing cancellation dates that no longer existed anywhere in the source, with no error and no warning. That is the part that needs measuring, not the download.
How do you decide whether a filing is still valid?
Our unique key is (USDOT, cancellation date, insurer), so when a date changes upstream the filing arrives as a new row and the old one is simply never touched again. Any lead the most recent ingest did not re-confirm is treated as unverified and withheld. That rule needs no knowledge of any particular incident and self-heals as the source recovers.
Can I cite these figures?
Yes. The measurements are ours, taken from public FMCSA records, and free to cite under CC BY 4.0. Please link back so readers can see the current figures rather than a snapshot.
Citing this
XDate Alert. FMCSA insurance feed status: daily observations of forward-dated cancellation filings, 2026-06-11 onward. Observation record v1.0, 2026-08-09. Licensed CC BY 4.0.
Available at: https://xdatealert.com/data/fmcsa-insurance-feed-status
The measurements are ours; the underlying filings are public U.S. government record. Reuse them, with attribution. Corrections and independent replications are welcome — we would rather be corrected than uncorroborated.
Sources
Related: cancellation trends by state and insurer · what happened to the FMCSA L&I system