Greenhouse Job Delta Monitor: Get Only the Roles You Haven't Seen

Point this at a list of Greenhouse job boards, put it on a schedule, and each run hands back only the postings that were not on the board the last time it looked. One row per new job, with title, location, posting date and apply link. It reads Greenhouse's public board API directly, so there is no login and no page scraping.

Who it's for

A revenue-operations analyst at a sales-tooling vendor. Your ideal customers are scaling SaaS companies, and a freshly opened "Head of Sales Operations" requisition is the best timing signal you will get. You keep 150 target accounts in a sheet, you do not want to reread 150 careers pages every Monday, and you do want a Slack message the day a matching role appears.

A researcher at a small executive-search firm. A venture fund hands you its portfolio list and asks which companies are hiring for finance leadership. You need the new roles each week, not the 700 that have been open since spring.

A founder watching three competitors. Hiring is the cheapest public view into where a rival is investing. A new cluster of machine-learning roles at one competitor tells you something no press release will.

All three need the same thing: the difference between this week and last week, delivered without re-sorting a full board.

Why this one

• You pay for new jobs, not for looking. Billing is per job returned. A run in which nothing changed returns no jobs and costs only the run start fee.
• Direct from the board API. It calls boards-api.greenhouse.io, the endpoint Greenhouse's own embeds use. A redesigned careers page does not break it.
• Your history lives in your account. The list of jobs already seen is a key-value store in the Apify account that runs the actor, not in ours.
• Safe against a spending cap. If you set a maximum cost per run and it is hit, the jobs that were not delivered stay marked as new, so the next run still returns them.

What you get

One row per newly detected posting. The columns are the same on every run:

Field	Meaning
job_id	Greenhouse's numeric posting id, as text. This is the key the baseline uses.
title	The job title as posted.
location	The location name on the posting, for example "San Francisco" or "Remote".
departments	Always an empty array in practice. See the note below.
posted_at	Greenhouse's first-published time, falling back to last-updated. Returned as Greenhouse formats it, with its own UTC offset.
url	The apply page on the company's Greenhouse board.
company_board_id	The board token the row came from.
first_seen_timestamp	When this monitor first recorded the posting.

About departments. Greenhouse's job-list endpoint does not include departments. On the four boards we checked on 2026-10-08 (stripe, airbnb, gitlab, figma) not one of 1,260 postings carried them, so the column is an empty list. This also means a keyword filter cannot match on department names. If you need departments, use the lookup actor listed under related actors; it makes the extra request this monitor skips.

How the baseline works and how to reset it

The monitor keeps one record per board. Each record maps a job id to the moment it was first seen.

1. Run one. The store has no record for the board, so every open posting that passes your keyword filters counts as new and is returned and billed. On Stripe's board the first run returned 726 jobs, which is about $0.73.
2. Run two onward. The monitor fetches the board again and returns only job ids missing from the record. A posting that is closed and then reposted under a new id comes back as new.
3. Where it is stored. In a named key-value store, greenhouse-delta-store unless you set deltaName. Inside it, the key is baseline-<board token>, for example baseline-stripe.
4. Keeping users apart. Two schedules with different deltaName values keep completely separate baselines, even for the same board. Use one name per person or per sheet.
5. Seeding without a bill. Turn on skipFirstRun and the first run only writes the baseline, limited to the postings your keyword filter lets through, and returns nothing, so you pay for new jobs from run two on.
6. Resetting. To make one board look brand new again, open Storage, then Key-value stores, open the store, and delete the baseline-<token> key. To reset every board, delete the store. The next run is a first run again.

Filters, caps and the baseline. The stored record lists only postings that were actually handed to you. A posting hidden by includeKeywords or excludeKeywords is never written to it, so widening the filter later delivers that posting as new on the next run. The same goes for postings cut by maxResults, postings left unbilled when a spending limit stops the run, and a whole batch whose write to the dataset failed: none of them is recorded, so none is lost. Reset is only needed when you want already delivered postings again.

Pricing

• Job posting returned: $1 per 1,000 job postings

Plus a $0.00005 start fee per run. Each event above is billed independently, only when it actually returns data — misses (found:false) are never charged.

Because the unit is the job, the first run is the expensive one. A 700-job board costs roughly $0.70 once, and a typical weekly delta of a handful of postings costs a fraction of a cent. Setting skipFirstRun removes the first-run charge entirely.

How to use

1. In the Apify Console. Open the actor page and click Start — the company_board_id field is already pre-filled with a working example. Results land in the run's dataset as soon as each item is found.
2. Via the API. Call it directly with a POST request — no Console needed once you have an API token:

   curl "https://api.apify.com/v2/acts/accountable_eel~greenhouse-job-delta-monitor/run-sync-get-dataset-items?token=<YOUR_TOKEN>" \
     -X POST \
     -H "Content-Type: application/json" \
     -d '{"company_board_id":["stripe"]}'
   

3. On a schedule. Save this actor as an Apify Task with the input you want, then add a Schedule (hourly, daily, weekly) so it runs on its own — no server of your own required.

Paste a bare token such as stripe, or the full careers URL. Both boards.greenhouse.io/stripe and job-boards.greenhouse.io/stripe are recognised.

Input

{
  "company_board_id": [
    "stripe"
  ]
}

Greenhouse board token or full URL (e.g. "stripe" or "https://boards.greenhouse.io/stripe"). Charged per job posting returned. Boards with no result are free. Accepted formats: stripe, https://boards.greenhouse.io/stripe.

The other settings are includeKeywords and excludeKeywords (case-insensitive, matched against title and location), maxResults (a cap on rows per run; postings beyond it stay new for the next run), testRun (only the first five boards), onlyFound (hide the rows for boards that could not be read), deltaMode (on by default; turn it off to return every job on every run), deltaName and skipFirstRun.

Sample output

query	found	status	job_id	title	location	departments	posted_at	url	company_board_id	first_seen_timestamp	scrapedAt
stripe	true	OK			<location>						1970-01-01T00:00:00.000Z

A board that does not exist comes back as one found: false row with the status NOT_FOUND. Any other API failure is reported with the status ERROR. Neither is billed.

Use it from Clay, n8n, Make, or an AI agent

This actor runs synchronously over plain HTTP — call it directly from a script, a workflow tool, or an AI agent, no Apify Console needed once you have an API token.

curl "https://api.apify.com/v2/acts/accountable_eel~greenhouse-job-delta-monitor/run-sync-get-dataset-items?token=<YOUR_TOKEN>" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{"company_board_id":["stripe"]}'

n8n. Add an HTTP Request node: Method POST, URL https://api.apify.com/v2/acts/accountable_eel~greenhouse-job-delta-monitor/run-sync-get-dataset-items?token=<YOUR_TOKEN>, Body Content Type JSON, JSON Body {"company_board_id":["stripe"]} (swap in an expression from an earlier node for a real value).

Clay. Add an "HTTP API" column: Method POST, URL https://api.apify.com/v2/acts/accountable_eel~greenhouse-job-delta-monitor/run-sync-get-dataset-items?token=<YOUR_TOKEN>, Body {"company_board_id":["{{job posting}}"]}, mapping the row's job posting into the company_board_id array.

MCP. In Claude, Cursor, or any MCP client with the Apify MCP server, ask for "Greenhouse Job Delta Monitor | Apify" — the agent will find and run this actor.

Tips

• An empty dataset is the normal result of a quiet run. There is no "nothing new" row. To tell a quiet run from a broken one, look at the run status and the final log line, which reports how many rows were emitted.
• Find the token in the careers URL. For boards.greenhouse.io/acme, the token is acme. It is usually the company name, but not always.
• Start with testRun. It limits the run to the first five boards while you confirm that your tokens resolve.
• Pipe the rows somewhere that alerts. The useful setup is a schedule, then a webhook into n8n or Make, then a Slack or email step that fires only when the dataset is non-empty.
• Mind the first run on big boards. Run a large list with skipFirstRun on, or accept a one-time charge of about a tenth of a cent per open posting.

vs. alternatives

	What it costs	What you get	Trade-off
This actor (greenhouse-job-delta-monitor)	$0.001 per job returned, $0.00005 per run, nothing for a quiet run	New postings only, with a first-seen timestamp, kept in your own store	Greenhouse only, no departments, and the first run bills the whole board unless you seed it
Greenhouse Jobs Lookup (greenhouse-jobs-lookup)	Billed per company, listed at $0.012 per board found	The whole board each time, with departments, richer filters and its own optional new-only mode	When your list is short or you want departments, this is the better choice. It also wins on a board that posts 12 or more new jobs between runs, because one board costs the same however many jobs it holds.
Greenhouse's own API	Free	The same JSON	You write the seen-list, the retries, the scheduling and the flattening yourself

Prices for the sibling actor are the figures on its page at the time of writing; check it before relying on the comparison.

FAQ

Why did my second run come back empty?

Because nothing new was posted. The monitor returns only job ids it has not recorded, and a board with no changes yields zero rows and no charge beyond the run start fee.

Does it tell me when a job is taken down?

No. It only reports additions. A closed posting simply stops appearing in the board response.

What happens if I add a company to the list later?

That board has no record yet, so its first run returns every open job. Add it with skipFirstRun on if you would rather not be billed for its backlog.

Can I run it without delta mode?

Yes. Set deltaMode to false and every run returns every job your filter keeps, billed in full, still stamped with the time each was first seen.

Related actors

• Greenhouse Jobs Lookup is the per-company lookup with departments and filters, for a snapshot rather than a feed.
• ATS Jobs Unified Lookup works out which applicant-tracking system a company uses when you do not know.