Google Jobs Scraper — salary, apply links and full job descriptions

Scrape Google Jobs listings in 18 countries and 7 languages, with the full job description, the parsed salary, the posting date, the employment type, the job highlights and the real application URLs (LinkedIn, Indeed, Glassdoor, the company's own careers page) in every row.

$0.003 per job delivered ($3 per 1,000). Searches that return nothing, duplicates, filtered-out jobs and error rows are never charged.

No login. No account. No cookies to paste.

What you can use this for

What arrives in a row

Every figure below was measured on 13 Oct 2026 on n = 439 unique jobs from 49 Google Jobs result pages, 33 searches, 15 countries and 5 languages. We do not list a field whose fill rate we have not measured.

Re-measured on the Apify platform on 14 Oct 2026: n = 2,085 jobs from 50 private test runs covering all 18 countries and 7 languages. A job row carries on average 34.1 filled fields (median 34) and 25.1 filled job-data fields (not counting the echo of your input and run metadata). Description 98.2% (full text 83.4%), resolved application URLs 99.5% with 4.92 options per job, employment type 97.2%, posting date 82%, salary 45% (US 51.7%, n = 1,221).

Always there — 100% (n = 439)

id · title · company · applyLinks · applyUrls · shareUrl · employmentTypes (97.5%) · location (99.8%)

The application links are the part nobody else resolves. Google wraps every "Apply on …" button in a redirect. This actor follows each one and gives you the real destination (applyUrl), the same URL without tracking parameters (applyUrlClean) and its domain (applyDomain). On average a job has 4.87 application options (n = 439).

Usually there

Salary depends on the country, so one blended number would mislead you. Measured share of jobs with a salary: 40.4% in the US (n = 203), 46.7% in Canada, 33.3% in the UK, 20% in Germany, 6.3% in India, 10% in South Africa.

Job highlights (qualifications, benefits, responsibilities) are rendered by Google only on English (US) pages: 91.1% of US jobs have them and 0% elsewhere (n = 439). In Spanish, French and German Google serves a ~400-character description preview instead of the full text; those rows say so with descriptionIsPreview: true.

The salary, parsed

"$140K–$200K a year" → salaryMin 140000, salaryMax 200000,

salaryCurrency "USD"

How it compares

Competitors measured on 13 Oct 2026 through the Apify API (their public stats and published output schemas; no third-party actor was run). This actor measured on 14 Oct 2026 in 50 private runs on the platform.

"Not declared" means the field does not exist in that actor's published output schema; we read the schemas, not the marketing. For competitors, "at most" is the number of fields they declare: a declared field can still be empty. 12 of the 20 Google Jobs actors with 5+ monthly users publish no output schema at all.

Every one of the 50 test runs charged exactly the rows it delivered (2,085 jobs, 2,085 charges) with 0 duplicate ids. The same job keeps the same id across runs: 167 of 167 ids seen in more than one run had the same title and company.

Input

Example:

{
  "queries": ["registered nurse", "data analyst"],
  "locations": ["Chicago", "Houston"],
  "country": "us",
  "language": "en",
  "maxJobsPerQuery": 30
}

Output example (one row, shortened)

{
  "rowType": "job",
  "id": "PIaBVaFi308wULA3AAAAAA==",
  "title": "Senior Software Engineer",
  "company": "Example Corp",
  "location": "Austin, TX",
  "via": "LinkedIn",
  "postedAt": "2026-10-12T14:00:00.000Z",
  "employmentTypes": ["full-time"],
  "salaryMin": 140000,
  "salaryMax": 200000,
  "salaryCurrency": "USD",
  "salaryPeriod": "year",
  "salarySource": "listing",
  "applyUrls": ["https://www.linkedin.com/jobs/view/0000000000", "https://careers.example.com/jobs/1234"],
  "applyDomains": ["linkedin.com", "careers.example.com"],
  "qualifications": ["5+ years of experience with distributed systems"],
  "description": "We are looking for…",
  "descriptionIsPreview": false,
  "isRemote": false,
  "isHybrid": true,
  "query": "software engineer",
  "searchLocation": "Austin",
  "country": "us",
  "language": "en"
}

Every row carries all 44 keys, with null where Google shows nothing, so CSV and spreadsheet exports keep their columns.

Searches that produce no data add one error row (rowType: "error") that explains what happened and what to try. Error rows are never charged. The ERRORS and RUN_SUMMARY records of the run's key-value store list them too.

Pricing

1,000 jobs cost $3. With deep pagination, 50 jobs from deeper pages cost 50 × $0.003 + one $0.20 session = $0.35. Duplicates within a run, filtered-out jobs, empty searches and error rows cost nothing. A run stops cleanly when it reaches the maximum cost you set.

Reliability and speed

• Google answers a Jobs search in about 20 seconds through the route this actor uses (median 19.8 and 21.1 s in two platform runs of 20 searches each, 13 Oct 2026). Searches run in parallel, rows are delivered as each page arrives and the status message updates every few seconds.
• The default input finishes in under two minutes: 8 to 82 s on the platform in 4 runs (14 Oct 2026, 10 jobs each). 500 jobs from 50 searches took 209 s at the default 512 MB.
• If Google pushes back, the actor retries with a fresh connection, then slows down, and always stops in time to finish inside your run timeout with the rows it already delivered.

Deep pagination (experimental)

Google only serves result pages 2+ to a browser session. With deepPagination on, the actor opens a short browser session to unlock them. In our platform test (13 Oct 2026) a working session gave 4 pages on 7 of 10 searches, but sessions could not always be started. When none can be started, the run keeps the first pages, says so in an error row (not charged) and finishes normally. Most users get more jobs per search with expandQueries and more locations instead.

FAQ

Do I need a Google account or cookies? No. The actor never logs in.

Is the job id stable? Yes. It is Google's own job id, the same across runs, so you can deduplicate or detect new listings between scheduled runs.

Are duplicates charged? Not within a run: a job found by two searches or two query variants is delivered and charged once. Between separate runs the same job can appear again, and it is charged again.

Why are some descriptions short? In Spanish, French and German Google shows only a preview of the description; descriptionIsPreview tells you which rows. Search in English (language: "en") for full text and job highlights.

What about personal data? Recruiter email addresses and phone numbers that appear inside descriptions are replaced with [email] and [phone].

Disclaimer

This actor extracts publicly available job listings shown by Google Jobs — no login, no account. You are responsible for how you use the data, including compliance with privacy laws and with the terms of the sites involved. It is not affiliated with or endorsed by Google.