The Employer Layoff-Risk Signal Reference, Source by Source

Before you sink a batch of applications into one target company, or before you sign its offer, you want to know one thing: is it quietly shrinking, or about to cut headcount. This is a lookup document for that question. It keys each public downside signal to the source it lives in, tells you what that source legally must disclose, how much lead time it gives you, and the blind spot that makes it lie, so you can jump to one row and leave with a defensible stable, watch, or avoid read.

This is not a listicle of vibes. "Your boss seems distant" is not a signal you can verify on a named employer in ten minutes. A WARN filing, a runway number, and a recruiter count are.

What each downside-risk source actually proves

There are four families of public signal, and they fire in a rough sequence: financial pressure first, then operational changes, then hiring posture, then the legal filing that closes the loop. Each family has a different lead time and a different way of misleading you.

The order is not strict. One career-advice source treats financial pressure as the opening move; some listicles put the hiring freeze first. Treat the stages as overlapping, and weight the signal by its lead time rather than its position in any list.

The load-bearing point of this whole reference: the earlier a signal appears in that flow, the more warning it buys you, and the later it is, the more reliable but the more useless it becomes. A WARN notice is the most reliable signal on the page and often the least actionable, because by the time it publishes you may have 60 days.

WARN thresholds by jurisdiction, and what escapes them

The Worker Adjustment and Retraining Notification Act is the one public record an employer is legally forced to create before a mass layoff, which makes it the spine of this reference and also its most treacherous source. Federal WARN covers employers with 100 or more employees, requires at least 60 calendar days advance written notice, and only fires when a plant closing or mass layoff hits 50 or more employees at a single site of employment.

That single-site, 50-person floor is the entire game. State mini-WARN laws move the floor, and where they do not, roughly 40 states default to the federal standard.

The federal mass-layoff test has two arms: 50 or more employees who also make up 33% or more of the site's workforce, or 500 or more regardless of percentage, during any 30-day period. California removed the 33% arm entirely and covers employers at 75. New York covers at 50 employees, demands 90 days, and can trigger at 25 affected workers. New Jersey now applies to any employer with 100 employees nationwide, regardless of full or part-time status, and adds severance of one week per year of service. Iowa, New Hampshire, New York, and Wisconsin can reach layoffs as small as 25 employees. Ohio's HB 96 took effect with a 50-employee threshold.

What never appears in this record is the calculated cut. Employers regularly terminate only 49 employees specifically to avoid the notification requirement. A 200-person company laying off 50 people spread across several offices, with no single site losing 50, files nothing under federal law. This is not a loophole an employer stumbles into; it is a design feature they size cuts around.

Aggregators: coverage versus reliability

Three public trackers cover layoffs, and they trade breadth against reliability in a predictable way. Crowdsourced trackers catch tech fast but miss everything else; government-sourced trackers are exhaustive but slower and US-only.

Layoffs.fyi, founded in 2020 by Roger Lee, tracks tech layoffs reported by media, announced by the company, or otherwise verified, and each entry carries a source link. It has aggregated over 450,000 layoffs and its data is free to use with attribution. Crunchbase updates weekly and pulls from media reports, its own reporting, social posts, and Layoffs.fyi, citing at least 127,000 US tech workers cut in 2025. A government-sourced tracker like WARNTracker holds around 8 million employees across 40,362 companies from 1988 to 2026, updates daily from official filings, and is more reliable for the US while missing smaller and non-US layoffs.

The mechanism behind the trade-off: coverage width comes from loose verification, and reliability comes from narrow official sourcing. You cannot buy both from one source, which is why steps 2 and 3 of the procedure make you check a government dashboard and an aggregator separately.

Reading lead time: the filing calendar is not the news cycle

A WARN notice is typically filed 60 days before the layoff event, but the gap between a public announcement and the published filing can run to 100 days, so you cannot assume the filing is your first warning or your last. Timing is where this record misleads most quietly.

In the documented Meta 2023 case, Meta filed its WARN notices with California on day 41, the state released them on day 56, employees went off payroll on day 100, and the company had announced the layoffs in its "Year of Efficiency" press release 100 days before the WARN notice was published. The press release served investors on the news cycle; the filing followed the legally timed 60-day worker calendar. Two clocks, one event.

The reverse also happens. Many times the WARN notice is the first public announcement of a layoff, and some of the layoff dates in trackers are set in the future. State databases add another wrinkle: Washington timestamps notices by the date its Employment Security department receives the notice, not the effective date of the layoff.

Private-company substitutes: runway and hiring pace

A private, VC-backed company has no filing duty, so runway age becomes the key proxy, and funding announcements are the least reliable version of it. Runway is the months of cash a company has before it must raise or cut.

A healthy startup runway typically spans 12 to 18 months, some extending to 22 to 24 months after a fundraise, and JPMorgan's 2024 review recommends 24 to 36 months for risk mitigation. For a candidate, anything under 12 to 15 months without a clear path to the next milestone is worth probing carefully. It is not a disqualifier, but the timing risk is real.

The trap is reading the headline raise as the runway. Funding announcements are marketing; runway is what matters. A large announced round can mask a short net runway if the company was already burning fast or raised into old debt. Ask for months of runway, not the size of the raise. Venture backing itself is a weak positive signal of employer quality, but it is not a substitute for the number.

Hiring posture: the signal that never files

Hiring capacity is measurable and it never appears in any legal filing, which makes it the most useful early read you can build yourself. Recruiter headcount tracks a company's aggregate hiring appetite, so a target employer with unusually few recruiters relative to its size is telling you something the filings cannot.

The baseline is lopsided by market. In Refolk's index of professional profiles, the United States holds 112,843 recruiter and talent-acquisition professionals against the United Kingdom's 7,342.

Use this as a hiring-capacity yardstick. A target US employer with a recruiter team that is shrinking, or tiny relative to its headcount, is showing you a hiring-posture flag before any freeze is announced. The same read applies to open requisitions: count them over time, and watch for freeze language in postings and careers pages.

Tracing who is on a company's talent team, and when each of them started, is exactly the kind of question a resume tool cannot answer but a people search can. Refolk writes your resume from your own history and tailors it per posting, and the same index that scores your fit lets you check whether a target's hiring team is growing or thinning before you commit a batch of applications to it.

The ten-minute lookup, step by step

Here is the procedure, ordered so each step either narrows the source families that apply or adds a dated signal to your read. Steps 4 through 7 overlap in time, so run them in whatever order the company's type makes easiest.

How this read goes wrong

Every source in this reference has a way of lying, and most bad reads come from trusting one source past its blind spot. These are the failure modes to check before you commit to a verdict.

"No WARN filing" read as safety. A 49-person cut, a multi-site cut with no single site losing 50, or a 60-person Texas firm all legally file nothing. Check headcount per site and the state threshold before concluding stability. The record is created by the same actor incentivized not to create it.

Effective date confused with filing date. A future-dated effective date makes a fresh filing look old. Read the received-date column separately.

WARN treated as the first alarm. If the WARN notice is your first public sign, you have roughly 60 days, not months. That is a decision window, not a warning.

Aggregator gap mistaken for absence. Layoffs.fyi is tech-only and crowdsourced. A non-tech or small private cut may simply be uncovered, so cross-reference a government tracker before reading silence as safety.

Hiring freeze over-read. A soft freeze is caution, not a verdict; freezes commonly run three to six months. Verify hard versus soft and ask for the review point.

Funding round age over-read. A big announced raise can mask a short net runway. Announcements are marketing; ask for months of runway, not the headline.

PIP mistaken for a layoff signal, or the reverse. A layoff hits peers and adjacent roles; a performance improvement plan is individual. Check whether documentation pressure is company-wide before treating it as a workforce signal.

Recruiter count misread. A high recruiter headcount can be staffing-agency noise rather than the employer hiring. Confirm the recruiters sit inside the target company, not a vendor. In Refolk's index, the top-count companies for such a query included staffing agencies, so the raw number needs that filter.

Assigning the verdict and keeping it current

Your final read is a one-line verdict backed by the two or three load-bearing facts, each cited to a source. The point is defensibility: you should be able to say why, not just what.

Weight the signals by lead time. A financial or leadership signal with 60 to 90 days of runway ahead of it is worth more than a WARN filing that leaves you 60 days, because the earlier signal gives you room to decide. Count independent signals rather than restating one signal three ways: a missed earnings call, a new CFO, and a recruiter team that halved are three signals; a press release, a news article about that press release, and a tweet about the article are one.

TARGET: [company], [public / private-VC / subsidiary], worksite state [state]
WARN: [filing found on date / none found; site + headcount + effective + received date]
HISTORY: [prior rounds since 2020 from aggregator, with source links]
FINANCIAL: [last earnings or last funding round + runway estimate, with date]
OPERATIONAL: [leadership change / reorg / closures, dated]
HIRING: [expanding / flat / frozen; recruiter count trend]
LEAD TIME: [approx. days of warning remaining]
VERDICT: [Stable / Watch / Avoid] because [signal 1], [signal 2], [signal 3]

Downside risk is a moving read, not a one-time score, so set it up to re-check itself. Washington's ESD database offers email alerts; subscribe for any state you would work in. Crunchbase refreshes weekly and government trackers refresh daily, so a monthly re-pull during an active search is enough for most targets and a weekly one is warranted for a company already in your "watch" corner.

Run this once end to end and it takes about an hour. Run it a second time on a different company and it takes twenty minutes, because you already know which source families apply and which columns lie. That speed is the point: it lets you clear a shortlist of targets before you spend a single application on the wrong one.