The Gaps in the Layoff Numbers: Undated Notices and Missing Worker Counts

Layoff Atlas currently shows 679 WARN notices covering 65,527 reported workers in its 90-day window. Those are the headline numbers, and they are useful. They are also incomplete in two specific ways that are worth understanding before quoting them: some notices have no published notice date, and some notices have no worker count at all.

Neither gap is a mistake in the data. Both are visible on the site itself, and both push the totals in the same direction: down. The real number of workers covered by recent WARN filings is higher than 65,527, not lower.

Gap one: notices with no date

The site's notice window is 90 days, and only notices with a published notice date enter that window. Records without a notice date appear separately, under "Recently imported," with their notice date listed as unavailable.

Five such records sit on the front page right now, covering 298 workers between them:

• Kenco Logistic Services, Myerstown, Pennsylvania: 81 workers, effective March 1, 2027
• Ask Foods, Palmyra, Pennsylvania: 96 workers, effective December 1, 2026
• Mars Information Services, Newark, New Jersey: 55 workers, effective January 31, 2027
• Republic Amusements, Camp Hill, Pennsylvania: 48 workers, effective December 20, 2026
• Starbucks, Philadelphia, Pennsylvania: 18 workers, effective December 6, 2026

These are reported plans, not confirmed job losses, and their effective dates are published — that is how we know when the affected workers are scheduled to be let go. But without a notice date, they cannot be placed inside the 90-day window, so none of those 298 workers are in the 65,527 total.

Contrast that with the dated filings at the top of the same page: Starbucks in Seatac, Washington, 135 workers effective December 31, 2026, with a notice date of October 23, 2026; Foster in Greenville, North Carolina, 12 workers effective March 5, 2027, noticed October 21; DSquared Hospitality in Bellevue, Washington, 83 workers effective December 20, 2026, also noticed October 21. A published notice date is what admits a filing into the window.

Gap two: notices with no worker count

The second gap is larger and harder to size. The site reports 33 counts unavailable inside the 90-day window — 33 notices for which no worker number was published. Worker totals exclude unavailable counts, so those 33 notices contribute zero workers to the 65,527 total even though each one represents a real filing affecting real workers.

There is no way to estimate those workers honestly from the published data. A notice with no count could cover a handful of workers or several hundred. That is exactly why the total should be read as a floor: it counts only the workers that notices actually reported.

What the totals still tell us

None of this makes the headline numbers meaningless. Within the 90-day window, the ranking of states is clear: Washington leads with 17,073 reported workers, followed by California with 14,001, Illinois with 4,626, Texas with 3,652, and Ohio with 3,180. And the upcoming weeks are heavy and sustained: 5,503 workers scheduled for the week of October 26, 5,061 for November 2, 6,568 for November 9, 6,145 for November 16, 6,120 for November 23, and 5,592 for November 30 — a six-week total of 34,989 reported workers.

Those upcoming figures are also reported plans, not confirmed job losses, and they carry the same caveats. Some of the workers in them come from filings whose counts are known; others will be added as missing counts surface or as undated records gain notice dates.

Why the gaps exist

WARN coverage is not a census of layoffs. The registry covers 51 jurisdictions, and 48 of them have imported records. Figures combine official records with labeled secondary reports, and reporting rules vary by state. Some states publish complete notices with dates and counts; others publish partial records, or records arrive through secondary sources that lack a field.

Missing source coverage is not zero layoffs, and a missing count is not zero workers. It is a blank field, and the honest way to handle a blank field is to leave it out of the total and say so — which is what the site does, with the "33 counts unavailable" note and the separate undated-records table.

How to read the numbers

When you see a Layoff Atlas total, read it with three qualifiers attached. First, it covers reported plans, not confirmed job losses. Second, it excludes notices with no published notice date, including the 298 workers in the five undated records listed above. Third, it excludes the workers behind 33 notices with no published count.

The result is a conservative number. If a future data refresh fills in those dates and counts, the totals will rise — not because layoffs suddenly increased, but because the picture got more complete. That is a feature of honest data, not a flaw in it.

Data source: Layoff Atlas