Refolk
September 4, 2026·9 min read

WARN Filings Fell 3.5% While Announced Cuts Jumped 118%. Read the Gap.

WARN notices trail 2025 by 3.5% while Challenger reports 118% more announced cuts. Here is what technical sourcers should mine instead in 2026.

WARN Act 2026 statisticssourcing laid off engineerstech layoffs 2026 vs 2025WARN notice sourcing playbookSB 617 Cal-WARN
WARN Filings Fell 3.5% While Announced Cuts Jumped 118%. Read the Gap.

The "pull the state WARN feed, cross-match to LinkedIn, blast the shortlist" playbook is quietly cooked in 2026. LayoffAlert's September update shows filings and affected headcount both trailing 2025, even as Challenger's January print screamed 118% more announced cuts than a year prior. Two signals moving in opposite directions is not noise. It is the market telling technical sourcers that WARN has become a lagging, narrower, and legally reshaped signal.

What the 2026 WARN numbers actually say

Through September 3, 2026, LayoffAlert.org tracked 3,245 WARN notices affecting 290,283 employees across 44 states, running 3.5% below 2025's same-period pace on notices and 3.8% below on affected workers. That is the entire foundation for the "layoffs are cooling" narrative, and it is misleading if you stop reading there.

The WARN Act (Worker Adjustment and Retraining Notification) is a federal statute that requires employers of 100+ to give 60 days' written notice before a mass layoff of 50 or more workers. California's Cal-WARN sets a lower bar: 75+ employees, same 50-worker trigger. Anything under those thresholds never appears in the feeds sourcers scrape.

MetricValueSource
2026 YTD WARN notices (through Sep 3)3,245LayoffAlert.org
2025 same-period WARN notices3,362LayoffAlert.org
YoY change, notices-3.5%derived
2026 YTD affected workers290,283LayoffAlert.org
2025 same-period affected workers301,891LayoffAlert.org
YoY change, workers-3.8%derived
Jan 2026 announced cuts (Challenger)~108,000Challenger, Gray & Christmas
Jan-to-Jan announced-cut change+118%Challenger
LayoffAlert historical notice base43,363+LayoffAlert.org
Historical affected workers6,049,773LayoffAlert.org

Seven weeks before the September snapshot, on July 15, the same tracker showed 2,891 notices and 263,079 affected workers. That is roughly 50 notices per week, a pace that will not close the gap on 2025 unless Q4 accelerates sharply.

Why announced cuts are up 118% while WARN is down

Because the composition of layoffs changed. AI-era restructurings favor rolling small-batch cuts that sit under the 50-worker federal trigger and never generate a filing. The result: WARN captures a shrinking fraction of the actual displacement event.

Three mechanisms are pulling the two curves apart:

  • Sub-threshold cuts. A 40-person "team consolidation" every three weeks across a 20-office footprint is invisible to WARN and dominant in press releases.
  • Attrition-shaped exits. Companies steering people into PIPs, "voluntary" separation packages, or internal-mobility limbo produce headlines without notices.
  • Contractor and vendor unwinds. Terminating a staffing agency's SOW moves hundreds of workers off a project with zero WARN obligation on the client.
118%
YoY increase in announced US job cuts, January 2026
Challenger, Gray & Christmas reported ~108,000 cuts in January 2026, the highest January since the pandemic, even as WARN filings ran 3.5% behind 2025.

A ResumeBuilder survey found 58% of companies plan layoffs in 2026, citing AI adoption, economic uncertainty, and restructuring. If even half of that is real, the ~3,245 WARN filers to date are a rounding error against the actual restructuring population. The playbook that reads the WARN feed and stops there is systematically missing the modal 2026 layoff.

SB 617 broke the California WARN feed for sourcers

California's SB 617, effective January 1, 2026, added four required disclosures to every Cal-WARN notice, and it made the feed noisier, not richer, for anyone matching candidates on filing dates. Signed October 1, 2025, the amendments apply to every notice issued on or after January 1, 2026, regardless of when the workforce reduction was planned.

Under SB 617, each Cal-WARN notice must now:

  1. State whether the employer will coordinate rapid-response services through the local Local Workforce Development Board (LWDB), another entity, or not at all.
  2. List the LWDB's direct contact information.
  3. Include standardized job-placement and retraining language.
  4. Confirm rapid-response services are arranged within 30 days after the 60-day notice, if coordination is planned.

A facially timely 60-day notice that omits any of the four is legally deficient. Each day of deficiency is a separate violation, back-pay liability is per employee, and the civil penalty runs daily. Every legacy template your favorite scraper is trained on is now non-compliant.

The predictable consequence: a wave of amended and refiled Cal-WARN notices in 2026. Anyone doing name-and-date matching against LinkedIn is going to double-contact the same engineer or, worse, contact the wrong cohort entirely.

The competition now ships with the notice

SB 617 also means the WARN feed literally advertises the competition. Every California notice publishes the LWDB's contact information and rapid-response service links. State workforce boards get the affected list first and must route displaced workers into placement services within 30 days.

Private recruiters showing up on day 60 with a cold LinkedIn InMail are late to a queue the state built. LWDBs are now a competitor category, and they have both the list and the mandate.

WARN is a Q3 2020 tool being run in a 2026 market, and the feed itself now advertises the competition.

Everyone is fishing the same six ponds

The companies dominating 2026 WARN filings are the same companies dominating the "open to work" pool, which means WARN-based shortlists overlap heavily with what any tier-1 recruiter is already working. Amazon, Meta, Verizon, Oracle, Tyson Foods, and Spirit Airlines appear prominently in 2026 filings per Newsweek's tracking. Oracle alone eliminated approximately 1,200 California positions in early 2026. Meta, Amazon, Block, Google, and Pinterest each announced substantial reductions with significant California headcount impact.

In Refolk's index of professional profiles, the top companies among US engineers currently broadcasting "open to work" are Amazon, Google, AWS, and ByteDance. That is the exact overlap: WARN-heavy employers and self-declaration-heavy employers are the same names. If your shortlist starts with a WARN filing and ends at LinkedIn, you are competing with every sourcer running the identical script.

The alpha is in tier-2 restructurings where competing sourcers are not running scripts:

  • Verizon telco and network engineering pools
  • Tyson Foods industrial software and supply-chain platform teams
  • Spirit Airlines operations tooling and revenue systems

These names do not trend on Hacker News. They also do not have 40 recruiters already in the inbox.

Why "open to work" is a rounding error

Because self-declaration is rare, and the ceiling on any signal that depends on it is tiny. In Refolk's index, about 555,684 US profiles currently hold a title of Software Engineer, Senior Software Engineer, or Staff Software Engineer. Of those, only about 199 carry "open to work" language in the headline.

3.6 per 10,000
US engineers broadcasting "open to work" at any moment
From Refolk's index: ~199 out of ~555,684 US software engineers signal availability in-headline. WARN-based shortlists compete for this tiny visible slice.

That is 3.6 engineers per 10,000. Whatever WARN-based enrichment you layer on top is competing for that same visible slice. The rest, meaning virtually all of them, surface only through pattern-of-life signals: sudden GitHub activity spikes, private repos going public, LinkedIn "About" edits, conference talk withdrawals, a stale portfolio site getting a Tuesday-afternoon refresh.

This is exactly the friction Refolk is built to remove. You describe the person in plain English (recent Oracle exit, backend Python, shipping side projects) and get a ranked shortlist across GitHub, LinkedIn, and the open web, without babysitting a WARN scraper or a boolean string.

What to source on instead in 2026

Stop sourcing on filings and start sourcing on the pattern-of-life signals that lead them by 30 to 60 days. Below is the practical replacement stack for the WARN notice sourcing playbook, ordered by lead time.

1. In-progress restructurings, not filed ones

Track the intent signals that precede a WARN notice by weeks:

  • Reorg memos leaking to The Information, Business Insider, and Blind
  • Internal recruiter LinkedIn headline changes ("open to opportunities" from someone still employed)
  • Manager-level departures clustering in the same org chart within a two-week window
  • Job requisitions quietly pulled from a company careers page

2. Tech-hub consolidations

Return-to-office and hub consolidations displace engineers who never appear on a WARN. When a company mandates Seattle or New York and an engineer lives in Austin, that is a departure event with no filing. Watch:

  • Real estate press releases about office closures
  • SEC 8-K filings citing "workforce optimization"
  • Company all-hands leaks announcing hub-city changes

3. Behavioral tells across GitHub and the open web

The signals that actually correlate with an engineer being reachable next Tuesday:

  • Un-privating a repo or bumping a commit streak after months of silence
  • Publishing a technical blog post for the first time in a year
  • Registering a personal domain or refreshing a portfolio site
  • Withdrawing from a conference talk or CFP
  • Starring a run of "hiring" or "who is hiring" repos on GitHub

These are the queries that used to require an analyst, a scraper, and a weekend. Ask Refolk in plain English ("senior ML engineers who left Meta after August 2026 and have pushed to a public repo in the last 30 days") and the ranked shortlist is the output. That is the replacement for recruiting after layoffs by WARN row.

4. LWDB partnerships in California

If you actually want the Cal-WARN cohort, get on the LWDB rapid-response distribution list before the 30-day window closes. Under SB 617 that list is now the first touchpoint, not the last. This is a phone call and an email, not a scraper.

FAQ

Are tech layoffs actually slowing in 2026 vs 2025?

No. Headline WARN filings are running 3.5% behind 2025 through September, but Challenger reported roughly 108,000 announced US job cuts in January 2026 alone, up 118% year over year and the highest January since the pandemic. The two data sources measure different things: WARN captures only cuts that hit the federal 100-employee, 50-worker threshold (or California's 75/50), while Challenger counts announcements regardless of size. The compositional shift toward rolling sub-threshold cuts explains the divergence, not a slowdown.

Is the WARN notice sourcing playbook still worth running?

It is worth running, but not as a primary channel. WARN is now a lagging signal that overlaps heavily with cohorts every tier-1 recruiter is already contacting, and in California SB 617 is producing amended and refiled notices that create duplicate rows. Use WARN as one input among many, weight it below behavioral signals like GitHub activity and profile edits, and pay attention to tier-2 filers (Verizon, Tyson, Spirit) where competing sourcers are not paying attention.

What did California's SB 617 change about Cal-WARN notices?

SB 617, effective January 1, 2026, requires every Cal-WARN notice to state whether the employer will coordinate rapid-response services through a Local Workforce Development Board, list the LWDB's contact information, include standardized job-placement and retraining language, and confirm services are arranged within 30 days of the 60-day notice period. Any notice missing one of the four is legally deficient, with per-day, per-employee back-pay exposure. For sourcers, this means noisier California data and a state workforce board that reaches displaced workers before private recruiters do.

How rare is "open to work" among US software engineers?

Extremely rare. In Refolk's index of about 555,684 US software, senior, and staff engineers, only roughly 199 currently carry "open to work" language in their headline, which is about 3.6 per 10,000, or 0.036%. Any sourcing method that depends on self-declaration is competing for that same tiny visible slice, which is why behavioral signals (GitHub pushes, portfolio refreshes, LinkedIn "About" edits) matter more than any single filing feed.

Try it on the search you came here for

Stop building boolean strings. Just describe the person.

Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.

  1. 01Describe them

    One plain sentence. Role, city, stack, stage, whatever matters to you.

  2. 02I read the web live

    GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.

  3. 03You read the shortlist

    Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.

  • No boolean, no filters, no seat to buy. One box.
  • Read at search time, so a profile updated yesterday counts today.
  • Every step visible as it runs, every name with its reason.

500 free credits on sign-up. No card, no demo call. See real searches.

Read next