Data sources / Jobs feed, hiring signals and posted pay
Jobs feed, hiring signals and posted pay: sources and method
One row per open job ad on an employer’s own career site, plus the daily counts and changes per company derived from them (open jobs, new countries, posted-pay share).
- Row
- Job ad; company-day for the signals
- Key
- Employer website domain (registrable domain)
- Refresh
- Boards read once or twice a day; signals daily; jobs history since 2026-10-21 in this database.
Where it comes from
| Source | How we read it | Licence or basis |
|---|---|---|
| Employer career sites on applicant tracking systems (Greenhouse, Lever, Ashby, Workday, SmartRecruiters, Workable, Personio, Teamtailor, Recruitee, Oracle Recruiting Cloud) | The public job-board pages and the endpoints those pages load for every visitor, no login | Job ads the employer publishes to be found; read at low request rates |
| Sweden: JobTech (Arbetsförmedlingen) JobStream | Official open API, anonymous | CC0 1.0 |
| Public job listing sites | Public listing pages, no login | Public job ads; employer-level fields only |
How the data is collected
- We keep a directory of employer job boards, each matched to the employer’s website domain; the match is checked (country, name, domain) every night.
- Every board is read at least daily. A job that disappears is closed with the day it was last seen, so openings and closures are dated.
- Signals (net change, new countries, departments, posted-pay share) are computed from those dated rows; under 28 days of history they are marked provisional.
- Posted pay is parsed from the ad, converted to EUR with a fixed monthly rate table, and grouped by ISCO-08 occupation.
Legal basis and collection rules
- Job advertisements are published by employers so that anyone can find and read them. We read them as a visitor does, without logging in, and link back to the original ad.
- Only pages and endpoints that answer an anonymous visitor: no login, no account, no credentials, tokens or keys of anyone else, no paywalled content.
- Before a source goes live we read its robots.txt and terms and record what they say. They are one of the signals we weigh for each source, together with whether the data is public without login and what the data is.
- Low request rates, only the pages and fields we need, and we slow down or stop when a site signals it.
Personal data
- Recruiter and hiring-manager names, emails and phone numbers are not stored; free text is filtered for emails and phone numbers.
- Employers whose name reads like a private person’s name are left out.
- Pay bands are published only for cells with enough postings and employers, so a single ad cannot be singled out.
Details: privacy policy.
Quality and limits
- Nightly employer-mapping checks, published on the data quality scorecard.
- Staffing agencies and duplicate boards are detected and labelled, so counts are per real employer.
- Coverage is the employers whose career site we know; a company without a public job board has no rows.
- Counts are open ads, not headcount.
How to get it
Removal requests
If you run a source we read, or a record is about you or your business, write to [email protected] with the source or the record (a domain or a URL is enough). We stop collecting it and remove it from the files, the API, the MCP tools and the data pages within 24 hours, and reply once it is done.