Greenhouse Hiring Analytics

Turn a list of known Greenhouse job boards into clean job-post exports and board-level hiring snapshots. Built for company research, recruiting operations and labor-market analysis where you need counts by department, location and your own role keywords without assembling them manually.

What you get

• One CSV-friendly dataset row per public job post, deduplicated within each board
• Board token, company name (when supplied), title, location, departments, offices, job URL, post/requisition IDs and source timestamps
• Optional readable job descriptions
• Per-board department/location counts and overlapping title-keyword role buckets
• Counts of currently-open exported posts by first-publication month and within a chosen lookback
• Explicit completeness, cap, error and unknown-date indicators so partial data stays visible

This is a current snapshot. It cannot measure total historical posting frequency, jobs filled, hiring growth or removals from one run. Save repeated snapshots yourself for longitudinal research; a missing post alone is not proof of closure. A post is not necessarily a distinct role, requisition or hire, and prospect/talent-pool posts may appear.

Quick start

Use a known board token, for example greenhouse, figma or airbnb. Supported URLs include https://job-boards.greenhouse.io/figma and https://boards.greenhouse.io/figma/jobs/123. Company career-domain URLs, embedded board URLs and EU-hosted boards are not supported. The Actor never requests input URLs directly: it extracts a validated token and uses only the fixed public Greenhouse API host.

{
  "boards": ["greenhouse", "figma", "airbnb"],
  "maxJobsPerBoard": 500,
  "maxTotalJobs": 2000,
  "includeDescriptions": false,
  "lookbackDays": 30,
  "roleBuckets": {
    "management": ["manager", "director"],
    "engineering": ["engineer", "developer"]
  }
}

The board list is required in the input form. The code fallback for empty input is the Greenhouse company's public board.

Where to find results

Open Dataset for individual jobs and export as CSV, Excel or JSON. Open the run's Key-value store → SUMMARY for all board-level analytics and coverage. The same report is also saved as OUTPUT.

Each summary contains reportedTotalPosts, receivedPosts, uniqueReceivedPosts, exportedPosts, completeSnapshot, status, counts, date coverage and scope notes. Status is complete, partial, failed, blocked or skipped. completeSnapshot requires matching API total and raw count, valid unique IDs, and no output cap. Unknown totals, duplicates and invalid rows prevent a completeness claim. Successful empty boards can be complete. Failed and skipped boards never receive a zero-job business interpretation.

Analytics count only exported posts. A board that hits a cap has partial statistics, even though the API's reported total is retained. Departments and role buckets may overlap, so those counts may sum above the number of posts. Role buckets use case-insensitive whole-word/phrase matches in titles; they are buyer-defined heuristics, not verified seniority, employment level or a validated frontline/management classification. No buckets are assumed by default.

firstPublishedAt is the source's first-publication timestamp when available. updatedAt is kept separately and is never used as a new-job date. Missing or invalid dates stay unknown; future dates are excluded from recent-publication counts. Publication-month counts concern currently-open exported posts only and include a separate future-date warning count. observedAt is the per-board processing observation time in UTC. isProspectPost is true only when the source explicitly returns a null internal job ID, false for a supplied non-null ID and null when absent.

Limits and access behavior

• 1–20 boards per run, sequential requests; duplicate tokens are removed
• 1–2,000 exported posts per board; 1–10,000 exported posts per run
• One jobs-list request per board with content=true to obtain department/office data, even when descriptions are not exported
• Full list responses are downloaded before output caps are applied; boards with more than 16 MiB of response data are stopped
• A 100-second collection budget, 15-second socket timeout, at least one second between requests and any stricter robots delay/rate
• Reads robots.txt first; unavailable or disallowing robots policy blocks collection
• Redirects and other HTTP errors are not followed/retried; HTTP 401, 403 or 429 halts all remaining boards on the API host; no authentication, proxies, stealth or CAPTCHA handling
• Each description is capped at 50,000 characters and marked when truncated

The Actor makes public GET requests only. It does not fetch applicants, resumes, application questions or private recruiting data, and does not submit applications. Optional descriptions are public employer-provided text and may contain ordinary contact information as part of that text. Unknown or removed boards commonly return 404. Find the correct board token on the company's public careers page.

Partial runs retain useful successful boards with clear errors for other boards. If every board fails, is blocked or is skipped, the run fails after storing the error summary. Keep the platform run timeout at least 120 seconds. A platform kill before output is saved can prevent final result storage.

Source and responsible use

Uses the documented Greenhouse Job Board API, whose public GET endpoints do not require authentication. Greenhouse is a third-party service; this Actor is independently maintained and is not affiliated with Greenhouse. Public accessibility does not grant unrestricted reuse: follow applicable site terms, privacy requirements and your intended-use obligations. Source fields and availability can change.

Validation

Unit tests cover token validation and host boundaries, input bounds, deduplication, caps and partial counts, date uncertainty, keyword overlap, description decoding, robots rules, redirects and no-retry behavior. Run python -m unittest -v from the project folder. See LIVE_CHECK.json in the source package for the dated three-board sample and environment note. Live board counts change over time. Deployment verified on Apify on October 4, 2026: build 0.0.1 (df7OSN3Uf7e9nprAq) succeeded. Direct-network cloud run egXvuHrPUoKJqZ3fd succeeded in approximately 9 seconds and exported all 18 Greenhouse posts, with complete coverage, no cap, no duplicates and no invalid rows. Its input used boards: ["greenhouse"], per-board and total caps of 50, descriptions disabled, 30-day lookback, and management/frontline keyword examples. Both dataset records and the SUMMARY key-value report were inspected. This cloud smoke covered one board; the separate local three-board API smoke exported 334 posts (Greenhouse 18, Figma 162, Airbnb 154) using the workspace outbound gateway. These are dated verification counts, not a promise about future counts.

The source ZIP is a backup of the implementation, tests and validation evidence. The Actor is already deployed; no ZIP upload is needed to run it.