Google Jobs Scraper — pay per job
Pricing
from $4.00 / 1,000 job rows
One clean row per job from Google’s Jobs tab: parsed location, ISO posted date, salary split out, every apply link, the job’s own Google link. USD 0.004 a row, no subscription. US, UK, Canada, India.
Where, as a place name; the canonical form "City,Region,Country" (such as "Houston,Texas,United States") is the most precise.
The Google country site to search: google.com, google.co.uk, google.ca or google.co.in. These are the sites that serve the Jobs tab through the proxy; Australia, Ireland, Germany and France do not (tested 2026-10-27).
Only jobs Google lists as posted within the last week or the last day; "any" takes Google's default list.
Google serves 10 jobs a page. Above 10, the actor also fetches the narrower date windows (past week, then past day) as further pages, so a search yields up to about 30 distinct jobs; more jobs come from more searches (other locations or query wordings).
A list of {"query", "location", "country"} objects. A job two searches both list is pushed once.
How many searches run at the same time; each search's own pages are fetched one after another.
"scrape" pushes job rows; "capture" saves every raw page fetched to the run's key-value store, with a manifest, for building the parser.
Which Google page scrape mode fetches: udm8 (the Jobs tab), udm8-uule (the Jobs tab with the location encoded), udm8-jbr (the Jobs tab with jbr=sep:0), ibp (the older jobs panel) or web (the ordinary results page).
Capture mode fetches the first page of every search in each of these variants.
[ "udm8", "udm8-uule", "udm8-jbr", "ibp", "web"]Capture mode also fetches page two of the first search in each variant, to learn whether paging works.
Capture mode also fetches the first job's own page in both URL forms, to learn whether a description costs a request per job.
Capture mode also fetches, for every search, one more first page per entry: the named variant's URL with these query parameters added, to test a filter form before any code depends on it. Entries are {"name", "variant", "params"}; the name goes into the saved page's key.
Capture mode also requests the Jobs tab's infinite-scroll continuation this many pages deep after each search's first udm8 page, in each known URL format, to learn whether the proxy serves more than the first 10 jobs. 0 fetches none.
The extra parameter pages and scroll continuations are fetched for this many leading searches only; 0 means every search.
GOOGLE_SERP (billed per page, plain HTTP to Google) or RESIDENTIAL (billed per GB, HTTPS).
Capture mode fetches the first search's first page once more through the RESIDENTIAL group in every variant, for comparison.