39,006 WARN Notices in October: The Sourcer's BLS-Blackout Edge

The Federal Reserve Bank of Cleveland just logged 39,006 WARN notices across 21 states in October, one of the highest monthly totals since the dataset started in 2006. The federal jobs report is still dark because of the shutdown. If you source for a living, that combination is not a crisis. It is the clearest sourcing signal you will get this year.

Why October's WARN surge matters more than the Challenger headline

October's WARN surge matters because it is the only timely, name-level, employer-level labor dataset still publishing on schedule while BLS goes dark. Challenger counted 153,074 announced layoffs in October, the worst month in 22 years, but Challenger gives you a company and a number. WARN gives you a company, a site, a headcount, and an effective date 60 days out.

The headline figures:

• 39,006 WARN notices across 21 states in October 2025 (Cleveland Fed).
• Only four prior months since 2006 ran hotter: 2008, 2009, 2020, and May 2025.
• 153,074 total announced layoffs in October (Challenger), up 175% year-over-year.
• 1,099,500 cuts year-to-date through October, 44% above all of 2024.
• Nearly 450 companies announced layoff plans in October alone, the highest of the year.
• Warehousing and tech made up just over half of October's cuts.
• AI was cited as the reason for roughly 20% of layoffs; cost-cutting for about 33%.

The last official BLS jobs report reflects August. September is missing. Even once the shutdown ends, BLS backfills take weeks. Economists are reading Challenger like it is a Fed release because they have nothing else. You should not be reading Challenger. You should be reading the 21 state WARN portals Cleveland Fed aggregates.

39,006

What WARN actually is, and the 25% gap you need to respect

WARN is a 25% dataset, not a 100% dataset, and that is the first thing to get straight. The Worker Adjustment and Retraining Notification Act requires employers with 100 or more full-time workers to give at least 60 days' written notice before a plant closure or mass layoff, and most states publish those notices on a daily or weekly basis with company names and site addresses. That 60-day written notice is the entire sourcing play.

Here is the math on the gap:

What WARN misses:

• Employers under 100 FTE.
• Layoffs in the 29 states Cleveland Fed does not aggregate.
• Distributed cuts that never cross a single-site threshold.
• Voluntary separations and performance exits.
• Attrition, hiring freezes, and quiet reorganizations.

Treat WARN as the fastest lead-gen list in recruiting, not as a market read. Pair it with Challenger for coverage, 8-K filings for the public-company color, and layoffs.fyi (Roger Lee's tracker) for the distributed tech slice WARN under-indexes.

The real asymmetry is 60 days, not 39,006

The headline number is a distraction. The real edge is that WARN layoffs lead state unemployment-insurance claims and lead changes in the unemployment rate and private employment, per the Cleveland Fed's own working paper. For a recruiter that means you can message an engineer 45 days before their LinkedIn headline says Open To Work.

In Refolk's index of professional profiles, 560,778 U.S.-based software engineers (including Senior and Staff) are indexed. Only 137 of them currently flag Open To Work in their headline. That is 0.02%. Every other laid-off engineer is invisible on the usual filters until their status changes, which usually happens on or after their effective date. WARN surfaces them before the change.

You can message an engineer 45 days before their LinkedIn headline says Open To Work.

The sourcing move is simple:

1. Pull a WARN filing on Monday.
2. Extract employer, site, headcount, effective date.
3. Build a target list of current employees at that employer, in that city, in the roles the filing implies.
4. Reach out before the effective date, framed around the specific closure.
5. Track the effective-date cohort as a batch; follow up at week two and week six.

A WARN that names 300 at a Fremont site is a 300-person target list before anyone else knows those 300 names.

Why warehouse and logistics WARN events hit harder than tech

A single UPS, Amazon, or Target WARN can represent 5 to 10% of the national mid-market talent pool for a logistics specialty in a week, because the pool is thin. In the Refolk index there are roughly 1,548 U.S. warehouse and logistics operations managers inside warehousing-industry companies. A 150-person Amazon FC closure is a measurable dent in that supply.

Software engineering does not work this way. 560,778 engineers absorb a 500-person cut without the market noticing. That is why most sourcers over-index on tech WARN and under-index on operations WARN. The dollars are in the thin pools:

• Warehouse operations managers.
• Last-mile logistics leads.
• Clinical trial managers at a single-site biotech.
• EHS and plant managers at a closing manufacturing site.
• Specialty nursing when a regional hospital system contracts.

When CBS lists Target, Amazon, and UPS as the recent anchors, do not read it as a tech story. Read it as a thin-pool arbitrage. The Cleveland Fed layoff tracker is a logistics recruiter's best friend right now.

The friction most sourcers hit on this workflow is translating a WARN PDF into a real target list: the filing names the employer and site, but you still have to find the 300 individuals who work there in the specific function you recruit for. That is the exact gap Refolk closes. You describe the person in plain English ("ops managers at the Amazon ONT9 fulfillment center, 3+ years tenure, not already in a new role") and get the ranked shortlist back.

The AI-cited cuts are not on WARN forms

Roughly 20% of October's layoffs were cited as AI-driven, and those cuts are systematically under-represented on WARN. The mechanism is structural: WARN's trigger is site-based (100+ FTE at a single employment site), and AI-driven reductions tend to be distributed white-collar cuts spread across remote workers and small satellite offices. A 300-person distributed product org cut hits Challenger and the 8-K; it never files WARN.

If your beat is AI-exposed roles (product managers, mid-level engineers, data analysts, copywriters, L1 support), WARN alone will leave you blind to the largest category of 2025 displacement. Pair it with:

• 8-K filings for public employers (search EDGAR full-text for "restructuring" and "workforce reduction").
• Challenger's monthly PDFs for sector shares.
• Layoffs.fyi for private tech, often with headcount and sometimes with lists.
• Internal signals: unusual LinkedIn tenure clusters ending in the same month, Blind threads, CWA Local filings.

The WARN pipeline is necessary. It is not sufficient.

How to build a weekly WARN pipeline in four hours

Build it once, run it in two hours a week. The output is a Monday list of named employers with effective dates, which you convert into a candidate shortlist by Thursday.

1. Subscribe to the four highest-volume state portals

These four cover the bulk of the Cleveland Fed panel's volume:

• California EDD WARN report (updated weekly, downloadable CSV).
• New York Department of Labor WARN notices (RSS available).
• Ohio JFS WARN notices (weekly).
• Texas Workforce Commission WARN list (updated as filed).

Add Illinois, Washington, Massachusetts, and New Jersey if your coverage requires it. Cleveland Fed's own tracker, maintained by Pawel Krolikowski and Kurt Lunsford, aggregates state WARN notices into a monthly panel updated twice a month, so cross-check your weekly scrape against their release.

2. Normalize the fields

Every WARN notice gives you some version of: employer, site address, number affected, effective date, union affiliation (sometimes), and a contact. Normalize these into a single schema. Add a "function guess" column based on employer type (an Amazon FC is fulfillment, a Workday HQ cut is engineering and product).

3. Expand each filing into a candidate list

For each row, you need the actual people. A boolean on LinkedIn gets you partway. Named GitHub orgs get you the engineers who contribute publicly. The open web (press releases, team pages, conference bios) fills gaps for sales, marketing, and operations. The faster you can go from a filing to a ranked list of individuals, the earlier your outreach lands. This is where WARN filings as a recruiting workflow stop being manual labor and start being a research problem a tool should own.

4. Cadence outreach to the effective date

The effective date is the pivot. Reach out at:

• Minus 45 days: soft intro, no ask, market intel exchange.
• Minus 14 days: specific role, specific hiring manager, calendar link.
• Plus 7 days: follow-up with references at your company who have taken similar jumps.
• Plus 30 days: last pass before they land somewhere else.

Response rates on this cadence beat cold outreach to employed candidates by a multiple, because you are the first person in their inbox with a specific option.

137

The window closes in Q1 2026

Treat the current environment as a 90-day window, not a new normal. The last official jobs report reflects August conditions, September is missing, and BLS backfills will take weeks after the shutdown resolves. Once BLS is current again, the information asymmetry collapses: everyone sees the same unemployment rate and the same JOLTS print on the same day.

State WARN sites keep publishing on their normal weekly cadence regardless. But the specific edge, where WARN is the only real-time, name-level signal competitors have, is finite. Likely gone by end of Q1 2026.

The recruiters who build the pipeline this month will keep it after the window closes, because WARN still leads official data by 60 days even in normal times. The recruiters who wait for the shutdown to end and the BLS to catch up will be reading the same numbers as everyone else.

FAQ

How often does the Cleveland Fed update its WARN tracker?

Twice a month. Pawel Krolikowski and Kurt Lunsford's team aggregates state WARN notices from 21 state portals into a monthly panel, with updates pushed roughly every two weeks. If you need faster, scrape the individual state sites directly: California EDD and New York DOL both refresh weekly, and Texas and Ohio update as filings come in.

Why not just use Challenger instead of WARN?

Challenger is an announcement count, not a candidate list. Challenger told you October was the worst month in 22 years with 153,074 cuts, which is useful for sector shares and macro context, but it does not give you employer-by-employer site addresses, effective dates, or union locals. WARN does. Use Challenger to decide which sectors to hunt in; use WARN to find the people.

What about layoffs that never hit WARN?

WARN captures roughly 25% of total announced layoffs. The missing 75% is sub-100-FTE employers, distributed workforce cuts under the single-site threshold, voluntary separations, and the 29 states Cleveland Fed does not aggregate. For public companies, 8-K filings fill much of the gap. For private tech, Roger Lee's layoffs.fyi is the de facto cross-reference. For sub-WARN cuts in unionized sectors, the local union filings (CWA, UAW, SAG-AFTRA) are often more complete than any employer-side source.

What is the single highest-leverage WARN filing to watch this month?

Any thin-pool filing. A 150-person logistics cut at UPS or Amazon moves the national warehouse-manager market more than a 500-person engineering cut at a hyperscaler moves the software-engineer market, because there are only about 1,548 U.S. warehouse and logistics operations managers in warehousing-industry companies in the index versus 560,778 software engineers. Rank your WARN queue by denominator, not by headline headcount.