WARN Ledger

How the data is made

Plain description of where the rows come from and what we change.

Who runs this

WARN Ledger is built and run by an AI assistant for Muhammad Hamza. The assistant downloads each agency's public list, runs scripts that clean and combine them, and rebuilds this site. There are no editors checking each row by hand, so errors are possible. If a row looks wrong, the source link on that row is the authority.

We make no claims about how many people use the site.

Where the data comes from

State labor agencies publish WARN notices in their own formats: spreadsheets, web tables and PDFs. We read the public pages with a bot that names itself, waits between requests and respects each site's robots.txt. If a site blocks automated access, we skip it and say so. We do not get around blocks.

The list of states, formats, dates and missing states is in the sources file.

What we change

  • Put every state in one schema with ISO dates.
  • Map each state's wording to closure, layoff or relocation. The original wording is kept.
  • Drop amendments, updates and rescissions of earlier notices, so a notice is not counted twice.
  • Drop exact duplicates, undated rows and rows dated in the future.
  • Keep only notices from October 2024 onward.

Nothing is estimated or filled in. Where a state does not publish a field, it is empty.

Limits

  • This is not a complete count of US layoffs. Only employers covered by the WARN Act file, and only the 19 states listed here are in the data.
  • This is not legal advice. The federal law is summarized from dol.gov on this page.
  • See the quality report for gaps and null rates.