Is there a Google Jobs API?

Why the indexing API is not what most people are looking for, and how to read Google for Jobs results instead.

Two different things share one name

Search for a Google Jobs API and you will find an indexing API in the results. It is almost certainly not what you want, and the confusion costs people an afternoon.

• The indexing API is for publishers. If you run a careers site and want your vacancies to appear in Google for Jobs, you mark them up and tell Google they exist. It pushes data in.
• Reading results is a different problem. If you want to see which vacancies Google for Jobs shows for a search, there is no official public API for that. It pulls data out, and Google does not offer that door.

Almost everyone searching for "Google Jobs API" wants the second one: market research, competitor hiring, salary benchmarks, or feeding a job board.

What that leaves

Three routes, in order of how long they survive contact with reality.

• Scrape it yourself. Possible, and it breaks for the usual reasons: the results block is rendered dynamically, it is heavily localised, and automated traffic from a server address meets challenges that a browser never sees.
• Use a provider. Someone else absorbs the breakage. Compare on how they price an empty result and whether they let you fix the location properly.
• Go to the source instead. Often the better answer. If you care about a specific set of employers, reading their own careers pages gives you cleaner and more complete data than the aggregator, and it does not disappear when the aggregator changes.

Pinning the location actually matters here

Job results are location-first. A search without an explicit location returns whatever the requesting address looks like, which for a server is usually a datacentre in a country you do not care about. Any dataset you build without pinning the location is not wrong in an obvious way, which is worse: it is quietly about the wrong place.

What we offer for both halves

For the aggregator: a results dataset billed per result page delivered, at $20.00 per 1,000 pages, which is two cents a page. For the source-first approach: a dataset that reads a company's own careers site, billed per job listed at $1.00 per 1,000.

If you are tracking a fixed list of employers, start with the second one. If you are measuring a whole market, you need the first.

The datasets behind this