Sources and coverage

Job data sources

We collect job data from over 353k different websites across the web. All these sources are crawled using publicly available data. We scrape some websites directly, while others are accessed indirectly through job board aggregations.

When we break down jobs by source, a single job can be associated with multiple sources (for example, a company career page and one or more job boards). Because we count a job once per source, the sum of jobs across all sources can be higher than the total number of unique jobs in our dataset.

We crawl continuously rather than on a fixed nightly batch, and each posting carries date_posted (when it went live at the source), discovered_at (when we found it), and closed_at — so you can always tell a genuine hiring change from a change in our collection. See freshness for the current discovery latency and how we measure it.

Below, you'll find a breakdown of our largest job data sources and their contributions. If you need the full list, contact us.

Known coverage limitations

No collection of job postings is a census of the labour market, and ours is no exception. These are the biases we know about, so you can judge whether they affect your segment before you build on the data.

We only see companies that publish job postings. A company that hires through referrals, or without ever publishing a role, is invisible to us — not underrepresented, absent. Roles filled through a staffing agency are not excluded on principle: if the agency itself publishes the posting (on its own site or a job board), it's crawled and appears in our data like any other posting. Coverage is therefore deepest for companies with active, public hiring, and for roles that are conventionally advertised. If you are studying hiring at large, remember you are studying published hiring.

Roles that are never syndicated can arrive late, or not at all. See why some jobs are discovered later for how syndication delays affect discovery timing.

We deduplicate aggressively, so we count fewer postings than providers who don't. When the same title from the same company reappears within 30 days, we keep the first posting and ignore the repost. That is deliberate — it stops the same underlying job being listed twice and being charged twice in credits — but it means our volume figures are not comparable with a feed that bills every duplicate and every repost. When you compare providers on volume, check whether the other number counts postings or records.

scraping_source tells you where we saw a job first, not where it was published. We store each job once, the first time we find it, so a role we discovered on Greenhouse before it reached LinkedIn will never list LinkedIn as its source even though it was posted there too. The field is a collection artifact and should not be read as a measure of a source's share of the market. See why we don't recommend filtering by scraping source.

Coverage is not uniform across countries and languages. We collect postings from 195+ countries in their local languages, but the depth of source coverage varies by market. We are strongest in North America, Europe, Australia, Japan, and India, and a country whose hiring runs through a national board we do not yet crawl will look thinner than these. Rather than ask you to trust an average, the statistics page breaks volume down by country so you can check the markets you actually care about.

What we don't claim. We publish no independently audited fill rate per field and no third-party coverage benchmark. Where a figure exists we publish it and say how it was measured; where one doesn't, we would rather say so than imply a precision we haven't verified.

Checking coverage for your own segment

The disclosures above are general; your segment is specific. You can measure our coverage of it yourself, for free and before spending anything: searches and result counts consume no credits — you only spend when you reveal data.

• Run a job search or company search with your real filters and read the total number of matches.
• Do the same through the API with include_total_results: true on the Jobs Search endpoint.
• Take ten companies you know are hiring and look them up in the company lookup — the fastest way to test coverage of a specific niche.

FAQS: Frequently asked questions