No Talent, No Alpha
Data Room
Every number behind our analysis, in one place.
The same files we write from
When we write that layoffs have barely moved, or that benefits are rising faster than wages, the figures come from here. Public data, collected from government sources and rebuilt every morning, with every source named and every gap stated.
-
Every dataset in full
Layoff filings, job openings, pay growth, AI adoption and state labor markets, each with the source, the coverage and the caveats written out.
-
Ask a question in plain English
Answers are built from the data in front of you and name the source and period behind every figure. Nothing is estimated.
-
Search, filter and export
Narrow any dataset down to what you need, then download the rows as a CSV.
Reading the datasets…
WARN filings
Employers with 100 or more staff must give 60 days written notice before cutting 50 or more people at one site. The notice goes to the state, not the press. We collect those notices daily, direct from state agencies where they publish in a readable form and via public trackers where they do not.
Loading the current feed…
The feed did not load. It may be a temporary network issue.
Jobs in the filings we are tracking
–
Filings tracked
–
States with filings
–
Largest single filing
–
- Layoff
- Closure
- Outlined: city not published
| Company | Location | Jobs | Type | Filed | Source |
|---|
Nothing matches those filters.
California, New York and Illinois publish through dashboards that cannot be read automatically, so their filings are under-represented. Multi-month history is still accumulating, so this is a rolling window rather than an archive.
Reported layoffs
Cuts that were reported in the press but have not shown up in a filing, or where the filing covers only part of the story. Each one is sourced and every entry carries the outlet that ran it.
A press report is not a filing. Where a company announced a global figure without breaking out the US, we say so in the entry rather than guessing.
Openings and turnover
How many jobs are open, how many people are being hired, how many are quitting, and how many are being let go. Quits are the closest thing there is to a confidence measure: people leave when they think they can.
Loading from the Bureau of Labor Statistics…
This dataset could not be loaded.
| Month | Openings | Hires | Quits | Layoffs |
|---|
Openings are a count in thousands. Hires, quits and layoffs are monthly rates as a percent of employment, which is why they sit on a separate chart rather than a second axis. The most recent month is usually preliminary and gets revised.
Who is actually using AI
The Census Bureau asks roughly 1.2 million businesses, every two weeks, whether they used AI in any business function in the last fortnight. It is the only recurring official measurement of AI adoption that exists. Not a forecast, not a survey of executives about the future: what firms say they are doing now.
Loading from the Census Bureau…
The Census Bureau API did not respond.
– of businesses in the leading sector use AI.
– of businesses in the trailing sector do.
Census changed this question in October 2025, from whether the business used AI “in producing goods or services” to “in any business function.” That is a large part of why published adoption appears to jump from single digits to around a fifth of firms. Figures here are on the current wording, so they are not comparable with anything published before that date.
Self-reported, and binary: a firm where one person uses a chatbot counts the same as one that has rebuilt a workflow around it. Adoption is not deployment, and neither is impact. Census labels this an experimental product. One sector code the survey returns is unnamed and is left out of the ranking.
AI in job postings
The share of job postings that mention AI, daily since 2019. Where the Census survey measures what firms say they use, this measures what they are willing to advertise for, which tends to move first.
Loading the posting series…
The posting series did not load.
– of US job postings mention AI.
| Month | Share of postings | Change |
|---|
Postings are advertised demand, not hiring. Duplicate listings, roles that never get filled and the habit of adding “AI” to an unchanged job description all move this line without anyone being employed. It is a share of Indeed postings, so it can rise while the absolute number of AI roles falls.
What an employee actually costs
The government prices an hour of work for you. Not just pay, but the whole bill: wages, paid leave, insurance, retirement and the taxes an employer has to pay on top. This is the number behind every total-compensation statement, and the answer to whether your benefits load is normal.
Loading the compensation series…
The compensation series did not load.
– an hour is what an average private-sector employee costs. – of that is pay.
Benefits are – of the total. In state and local government the same hour costs –.
| Quarter | Total per hour | Wages | Benefits | Benefits share |
|---|
These are national averages across all employers, so they set the reference point rather than your number. Cost per hour worked counts paid leave as a cost but not as an hour, which is why the benefits share looks higher than a simple salary calculation suggests. The legally required line is derived as the remainder after the four published components.
What this cannot tell you: what to pay a specific job, what your competitors pay, or how to set salary bands. Public data has no concept of job level, and the percentile of an occupation is not a seniority ladder. Those questions need a paid survey, and anyone telling you otherwise is selling something.
How fast pay is rising
The Employment Cost Index is the cleanest read there is on the price of labor, because it holds the mix of jobs constant. When it moves, the cost of the same work moved, not the composition of who is working.
Loading…
This series did not load.
– is how fast private-sector pay is rising.
Benefits are rising faster, at –. The peak was – in –.
| Quarter | Private wages | Private benefits | All civilian | State and local |
|---|
This is not a merit budget. The Employment Cost Index measures what actually happened to the cost of a fixed set of jobs, holding the mix of occupations and industries constant. A merit budget is a forward-looking plan, and the surveys that publish those numbers are paid. Use this as the answer to "what did the market actually do", which is the harder question to argue with.
Who is adding jobs and who is cutting
Payroll employment by industry, and how each one has moved over the last twelve months. This is the answer to whether the pressure you are feeling is your company, your industry, or the whole economy.
Loading…
This series did not load.
– is growing fastest, at –.
– is shrinking fastest, at –. All employment together moved –.
| Industry | Employed (thousands) | 12-month change |
|---|
Temporary help is on its own chart because staffing turns before the wider market does, in both directions, which makes it the closest thing the monthly data has to an early warning. It bottomed in December and has risen since. Information has fallen every year since 2023, and it is the industry where the AI claims are loudest, so treat any single explanation of that line with suspicion.
State labor markets
Unemployment by state, and how it has shifted over the year. Hiring difficulty is local long before it is national, and the spread between the tightest and loosest state is usually wider than people assume.
Loading…
This series did not load.
– has the tightest labor market in the country at –.
– has the loosest at –. The biggest move over the year was –, –.
| State | Rate | A year ago | Change |
|---|
A low rate means it is hard to hire, not that the economy is healthy, and the two get confused constantly. The rate also counts only people actively looking, so a state can look tight because people stopped searching. Read it alongside the twelve-month change rather than on its own.
Work stoppages
Strikes and lockouts involving 1,000 or more workers, updated monthly. Industrial relations rarely feature in workforce data products, which is odd, because a stoppage is the most expensive people event most employers will ever have.
Loading…
This series did not load.
There have been – major work stoppages this year through –, involving – workers.
At the same point last year: – stoppages and – workers.
Over the last twelve months – workers have been involved in a stoppage, down from a peak of – in –.
| Full year | Stoppages | Workers | Days idle |
|---|
This counts stoppages involving 1,000 or more workers only, so it misses smaller disputes entirely and undercounts activity in fragmented industries. It also counts lockouts alongside strikes. The charts are trailing twelve-month totals rather than raw monthly counts, because at this scale a single large stoppage swings a month and tells you nothing about direction. The table underneath is complete calendar years.