Verification
Check our data before you rely on it.
Look up an employer to see how many of their jobs we hold and how new they are. Paste your own list to see which of your jobs we also have open, and which may have closed. Every result links to the original posting, so you can check it yourself.
The checks on this page are free and need no account. They read the same live data the Jobs API serves, and every result says how many jobs it looked at.
Check an employer
Type an employer's name. We count their current jobs, show how many are new and how many have been open for over a month, and list ten of them with links to the original postings.
Try , or .
Compare your list with ours
Paste up to 20 jobs from your own data, one per line as company, title, location, or upload a CSV with a header row. We look each one up in our current jobs and tell you what we found. If your file has a date column, we also show how old your copy is.
Your list is not stored. Each row's company and title words are sent as searches, and nothing else leaves your browser.
Header names we recognise: company or employer, title, location, url, and a date such as date_posted or last_seen. Rows that have only a URL need the URL matching in compare_sources.
Go deeper with an API key
Two tools in the Headcount Labs MCP connector run the full checks: one opens our jobs' original pages and compares them field by field, the other matches your whole list by posting URL. Ask your agent in plain words, or call the endpoint yourself. Both count toward your plan's monthly job allowance, like search results.
Spot-check us: verify_jobs
Picks up to 50 random current jobs, optionally from one employer or place, opens each original posting once and reports for every job:
- whether the link works and the job is still open
- whether the title, employer, location and pay match the page
- a summary with the percent passed for each check, out of the jobs where that check could be made
Ask your agent
Use Headcount Labs verify_jobs to spot-check 20 random jobs at Ford Motor Company. Show me the summary, then every job where a check failed, with its link.
Or call it
Compare your full list: compare_sources
Takes up to 500 rows per call, as rows or as CSV text, and matches by posting URL first, then by company, title and location. Every row gets a status:
have_current,have_closedorhave_unpublishedmissingwith a likely reason:company_not_covered,site_not_coveredornot_found_on_covered_board- per company, how many current jobs we hold that your list does not
Ask your agent
Compare the attached jobs file with Headcount Labs using compare_sources. Tell me what share of my rows you have, which of mine have closed, and the likely reason for each one you are missing.
Or call it
Keep your key in an environment variable such as HEADCOUNT_API_KEY, never in a prompt or a shared file. A check larger than your confirmation level asks before it runs. Set up the connector or read the API documentation.
How to read the results
These checks are built to show you where our data is weak as readily as where it is strong.
- 01
A sample is not a census
Ten jobs from an employer, or twenty rows from your file, tell you about those jobs. Every result states its sample size. Run a few checks on the employers and places you care about before you draw a conclusion.
- 02
Not found is not the same as wrong
A job may be newer than our last read of that board, may have closed before we saw it, or may sit on a board we do not read. On this page "not found" means we found no open match, nothing more. compare_sources gives the likely reason per row.
- 03
We count what we could not check
Some sites block automated visits or build their pages in the browser. verify_jobs marks those checks as not made and leaves them out of the percent passed, so a blocked site never counts as a pass.
Questions
Where do the numbers on this page come from?
From the live Jobs API, through the same no-account access our MCP connector offers to assistants. Each result shows the time of the data snapshot it read. Counts include only jobs we currently show as open.
Why is my count for an employer different from yours?
Three common reasons. We merge the same job posted on several boards into one job, so we often count fewer than a list of raw postings. We count only jobs that are open now. And the employer search matches names that contain what you typed, so a short name can include other employers; the result says when that happens.
What does "last read" mean?
It is the last time our reader finished reading the board that lists that posting. Some postings do not record it yet, and those show "not recorded". The coverage page shows how many boards the latest full pass read.
Do you keep the list I paste or upload?
No. The file is read in your browser. For each row, the company name and title words are sent as a search, the same as typing them into a search box, and the list itself is not saved.
I found a job that is wrong or closed. What should I do?
Use Report or remove a job. Reports go to a person, and closed jobs drop out of the data once our next read of that board confirms it.
Check it at full scale.
Get an API key to run verify_jobs and compare_sources on your own data.