Why Layoff Totals Can Fall Even as New WARN Notices Arrive

Layoff Atlas currently shows 704 WARN notices covering 66,156 reported workers. That total is lower than it was a week ago, even though new layoff notices kept arriving in the meantime. That apparent contradiction is not an error. It is how a rolling 90-day window works, and understanding it matters for anyone reading layoff numbers in the news.

The headline figures describe a notice window, not all layoffs ever recorded. Only notices with a published notice date inside the last 90 days enter the window; records without a usable notice date appear separately, under Recently imported. Every day the window moves forward: new notices enter, and notices older than 90 days drop out. The total can therefore fall even on a day with several new filings, if more old notices aged out than new ones came in.

The last week shows this clearly. On October 19, the window held 721 notices covering 66,487 reported workers. By October 25, it held 704 notices covering 66,156 workers — 17 fewer notices and 331 fewer reported workers. Yet new filings arrived throughout that stretch. Foster filed a 169-worker closure notice for Winston-Salem, North Carolina, on October 21, with a companion 12-worker filing in Greenville. DSquared Hospitality filed an 83-worker closure notice for Bellevue, Washington, on October 21. The Andersons filed a 64-worker closure notice for Montgomery, Alabama, on October 20. Starbucks filed a 135-worker notice for Seatac, Washington, on October 23. Those notices entered the window. The total still fell, which means at least that many older notices — and then some — reached the end of their 90 days and rolled out.

A single-day move makes the mechanism plainer still. The window stood at 737 notices and 67,101 workers on October 18 and at 721 notices and 66,487 workers on October 19. A drop of 16 notices in one day does not mean 16 employers cancelled layoffs overnight. It means the window's back edge moved past a batch of notices filed in mid-July, while that day's new notices did not fully replace them.

A falling window total is not a falling layoff trend

It is tempting to read a smaller total as good news — fewer layoffs — or a larger total as a worsening jobs market. Neither reading is safe on its own. The window total mixes two different flows: notices coming in and notices ageing out. To judge whether layoff activity itself is rising or falling, you need to look at the inflow as well as the total.

Three checks help. First, look at the newest notices themselves: the homepage's recent filings list shows what actually arrived in the last few days. A week with several large new filings is a busy week for announcements even if the window total dipped.

Second, look at the effective-date calendar, which is counted differently. Upcoming layoffs are grouped by the week they take effect, not by when they were filed. Over the next six weeks, 35,222 reported workers are scheduled to be affected: 5,206 in the week of October 25, 4,565 in the week of November 1, 5,311 in the week of November 8, 7,781 in the week of November 15, 6,527 in the week of November 22, and 5,832 in the week of November 29. The week of November 15 is the heaviest of the six. Because effective-date figures do not roll off like the notice window, they give a steadier view of what is already scheduled.

Third, compare like with like. A state page's calendar-year archive totals are a different measure from the 90-day window: Washington's archive shows 129 notices and 27,943 workers so far in 2026, California's 1,164 notices and 58,985 workers. Those year-to-date figures grow through the year and are not directly comparable to the homepage window total.

What the current numbers do — and do not — say

The current window covers 704 notices, 66,156 reported workers, and 33 notices with worker counts unavailable. Those unavailable counts are excluded from the worker total, so the true reported figure for the window is slightly higher than 66,156 by an unknown amount. Washington leads the state ranking in the window with 17,073 reported workers, followed by California with 14,224, Illinois with 4,626, Texas with 3,652, and Ohio with 3,180.

All of these figures describe reported plans, not confirmed job losses. A WARN notice is an advance filing an employer makes before a qualifying mass layoff or plant closure, and the final number affected can differ from the filed figure if plans change. Coverage is also partial: the registry covers 51 jurisdictions, 48 of which have imported records, and reporting rules and publication speed vary by state. A state with no recent notices in the registry has not necessarily had no layoffs; it may simply have no connected source publishing them yet.

So when the total rises, ask whether a large new filing arrived or whether few old notices aged out that day. When it falls, ask whether old notices rolled off — and check the new-notices list and the upcoming effective weeks before concluding anything about the direction of layoffs. The window total is a snapshot of the last 90 days of filings, and it changes partly because the frame itself moved.

Data source: Layoff Atlas