Job-board scraping vs. ATS source signals

Scraping job boards collects postings after they have been syndicated and, usually, after aggregators have re-listed them. That makes it easy and broad, but it inherits the layer's defects: the same req appears many times under slightly different titles and dates, staffing-agency re-posts are mixed in with direct employers, and the original requisition — the thing you actually want to count — has to be reconstructed from the copies.

Reading a req from the employer's applicant-tracking system or careers site collects it before syndication distorts it: one row per real requisition, attributed to the company that opened it, dated when the company opened it. These are ground-truth reqs, and they are what make a req-level buying signal specific enough to act on rather than merely count.

Board scraping wins on breadth and cost, and for a coarse market-sizing view — roughly how much hiring is happening in a sector — it is often enough. Source signals win whenever the use case depends on getting the company, the count, and the date right: competitive hiring intelligence on a named watchlist, account-level prospecting, or any dataset that has to be auditable.

In practice a serious hiring-demand dataset resolves board and aggregator copies back toward the source req rather than choosing one surface outright — deduplicating to one signal per company, role and country so the same demand is never counted twice. The landscape overview covers where each surface sits on that path.