535,557 distinct employers, and no monopolist. What looks like a giant hiring machine is mostly the same ad reposted across boards.
There is a story the job market tells about itself. A handful of enormous employers, a few platforms, and everybody else queuing up to apply. The platform half of that story is true. The employer half is not.
We cleaned and counted every employer name in the October 2026 edition, the full corpus rather than a sample, measured October 6, 2026. The result is 535,557 distinct employers across 7,477,415 postings. The 20 largest of them account for 4.6% of everything posted. The Herfindahl index across the whole market is 2.7, on a scale where 10,000 means one employer holds every ad.
That is about as close to perfectly atomised as a market gets.
If hiring is that diffuse, why does every job board feel like the same twenty companies?
Part of the answer is duplication, and it is measurable. 7.7% of the records we hold are a posting we already have under another source. A role that appears on the employer's career page, on two aggregator boards and in a staffing agency's feed is one job and four records. Our index does not merge them. We would rather show you both than hide a listing behind a guess about which copy is real.
The extreme version of this is not an employer at all. The most duplicated single record in the corpus belongs to a truck-driving board, not to a company with thousands of openings. That is one ad, syndicated. We publish the offending records on the hiring market page rather than describe them in the abstract.
5.9% of postings come from staffing agencies rather than the employer named on the ad. They are also the ones who put a figure in it: 20.3% of agency postings state pay, against 11.5% across the market.
Read that carefully. It does not mean agencies are more virtuous. An agency ad sits on the same board as everyone else's and has to win the same click, and a range is the cheapest way to win it. The employer posting on its own careers page already has the audience. Nothing forces the number out of them, so it stays in.
Three limits, and each of them pushes the concentration figure down rather than up.
Company names are cleaned, not perfect. "Walmart" and "Wal-Mart Stores, Inc." have to become one employer for this count to mean anything. Our cleaning gets the large cases right and will merge a few names that should have stayed apart. Every merge lowers the employer count and raises the top-20 share, so a sloppier job here would have produced a more dramatic number. We would rather have the boring one.
The corpus is not the economy. We index what public applicant-tracking systems and company career pages publish, which leans retail, healthcare, logistics, driving and skilled trades. An industry that hires through referrals, unions or a paper form at the door is thin here. This is a sample of what employers publish, and publishing is not the same as hiring.
Duplicates are counted as records. One employer with one ad on four boards counts four times, which inflates the tail and the head at once. We report that as 7.7% duplication rather than quietly deduplicating and publishing a number nobody could reproduce.
Employer concentration is a real research subject, and it uses this same index. Azar, Marinescu and Steinbaum compute it from online vacancy data and find that concentrated employers pay less; the NBER digest summarises the wage result.
Our number points the other way for employers taken nationally, and that is not a contradiction. Their unit is an occupation inside a commuting zone, and a single hospital can dominate the market for nurses in one town even while the national employer count runs to six figures. Concentration is local and occupational. Ours is national and across everything, because that is what one crawl can hold. Both can be true, and only one of them fits in a single number. We report the one we can measure and say which it is.
The size of a job board's inventory measures reposting, not hiring. When a board advertises millions of openings, the honest question is how many distinct roles sit behind them, and how many of those are still being served at the source. That second question is the one our crawl archive can answer, because we hold two dates for every posting: the one the source reports, and the one our crawler last saw.
The employer side, meanwhile, is a long tail with no head worth fearing. The interesting concentration in this market is not in who is hiring. It is in who is publishing the same job four times.
→ Hiring market structure · Where postings come from · Key figures · How we measure
Our numbers below are ours; these are other people looking at the same question. We paraphrase and link, we do not quote.
Want a cut we have not published? A region, an industry, a role family, a time series, one country against another — we can pull it from the same corpus these posts are measured on.
Tell us what you are looking at: [email protected]. We answer either way.
Source & licence. All figures here come from our own crawl of public job postings, measured against the corpus published in the monthly report and its JSON (CC BY 4.0).
Cite as: DeUnemployment, “The top 20 employers hold 4.6% of every job ad we measured”, https://deunemployment.com/blog/who-controls-hiring (data as of 2026-10-24).