LinkedIn Ghost Job Detector & Checker - Apply Link Verifier

Recover exact company application links and stop unsafe job rows before they reach your users.

LinkedIn job feeds often return an empty external apply URL, a copy of the LinkedIn job URL, or a generic company careers page. This Actor takes the job rows you already have, reconciles each one against public employer and ATS evidence, and returns a safe action for your workflow.

It is built for job boards, recruiting products, staffing automations, and job data pipelines. It is not another job scraper.

What it fixes

Feed problem	Actor decision
External apply URL is missing	Finds and proves the exact employer or ATS route
Apply URL opens a generic careers board	Resolves the board to the exact job
LinkedIn row remains visible after the official job closes	Returns EXPIRED with no publishable URL
Company, title, location, or job content conflicts	Returns SOURCE_MISMATCH or STATUS_CONFLICT
The job only supports LinkedIn Easy Apply	Returns LINKEDIN_EASY_APPLY and the LinkedIn route
Evidence is not strong enough	Returns AMBIGUOUS, publishes no URL, and does not charge

The safety rule is simple: no proof, no link.

Try it in 60 seconds

1. Keep the two prefilled example jobs.
2. Click Start.
3. Open the Verified apply routes dataset view.
4. Use actionUrl only when safeToPublish is true.

The prefill demonstrates both supported outcomes: an exact company-owned application page and a LinkedIn Easy Apply route. No LinkedIn account, cookies, or browser profile are required.

Input

Use either an existing Apify dataset or paste rows directly. Common LinkedIn scraper field names are detected automatically.

Each row needs:

• LinkedIn job URL or job ID
• job title
• company
• location

An existing applyUrl, companyApplyUrl, or company website is optional. It is treated as a hint and must pass the same verification policy.

{
  "rows": [
    {
      "jobId": "4441060956",
      "jobUrl": "https://www.linkedin.com/jobs/view/4441060956",
      "jobTitle": "Account Executive - Travel",
      "companyName": "Nike Communications, Inc.",
      "location": "New York, NY"
    }
  ],
  "maxItems": 25
}

For an existing dataset, provide its ID instead:

{
  "datasetId": "YOUR_DATASET_ID",
  "maxItems": 500
}

Output

Every input row produces an auditable decision. A verified result looks like:

{
  "status": "ACTIVE",
  "actionUrl": "https://nikecommunicationsinc.applytojob.com/apply/eDPh7BCIhx/Account-Executive-Travel",
  "safeToPublish": true,
  "reviewRequired": false,
  "usefulClassification": true,
  "confidence": 1,
  "sourceProvider": "jazzhr",
  "evidence": [
    "linkedin_apply_method:OFFSITE",
    "official_site_to_career_source_verified",
    "generic_board_resolved_to_exact_job",
    "conservative_official_inventory_match"
  ]
}

Important output fields:

Field	Use
actionUrl	Route your product may open or publish
safeToPublish	Machine-readable release gate
status	Final decision for the job row
reviewRequired	Whether the row should enter a manual queue
confidence	Strength of the match
sourceProvider	Verified ATS or official source
evidence	Checks supporting the decision
requestAccounting	Requests used by each verification layer
original	Original input row, unchanged

When the official job is gone

When the employer has removed a posting, the Actor says so instead of guessing. This record comes from a real run; values are exactly as returned, only the long evidence array is shortened here. It shows three guarantees at once:

• No fabricated URLs. actionUrl and officialUrl are null because no safe apply page exists. A missing value is always null, never a guessed link.
• Deliberate lifecycle status. EXPIRED is declared only after the complete official inventory was checked, as the evidence trail shows.
• Publish protection built in. safeToPublish flips to false, so a job board or alert email never ships a dead link.

{
  "jobId": "4415520094",
  "title": "Staff/Sr. Software Engineer, Search Data Infrastructure - Slack",
  "company": "Slack",
  "location": "Seattle, WA",
  "linkedinUrl": "https://www.linkedin.com/jobs/view/4415520094",
  "status": "EXPIRED",
  "actionUrl": null,
  "officialUrl": null,
  "applyMethod": "OFFSITE",
  "safeToPublish": false,
  "reviewRequired": false,
  "confidence": 0.95,
  "sourceProvider": "workday",
  "evidence": [
    "verified_company_website_map",
    "ats_link_on_career_page",
    "workday_cxs_api:query=staff sr software engineer search data infrastructure slack",
    "official_inventory_empty",
    "official_sitemaps_scanned:14",
    "complete_employer_owned_inventory_checked",
    "complete_official_inventory_no_safe_match",
    "expired_classification_without_url"
  ],
  "checkedAt": "2026-07-28T14:34:29.121893+00:00",
  "schemaVersion": "1.0"
}

Decision statuses

Status	Meaning
ACTIVE	Exact live official job route verified
AUTHORIZED_APPLY_ROUTE	Valid source-owned application route verified
LINKEDIN_EASY_APPLY	No external route is expected; use the LinkedIn URL
EXPIRED	The exact official job is no longer available
SOURCE_MISMATCH	The row appears to belong to a different hiring organization
STATUS_CONFLICT	Trusted sources disagree about the current state
AMBIGUOUS	Evidence is insufficient; no URL is published and the row is not charged

Why this is different

Most LinkedIn job Actors collect listings. Official career-site APIs sell their own job inventory. This Actor works between those products and your publishing system:

1. It preserves your existing job row.
2. It checks customer-supplied links instead of trusting them.
3. It resolves generic ATS boards to the exact job.
4. It compares LinkedIn, employer, ATS, and live detail-page evidence.
5. It blocks expired, mismatched, conflicting, and unproven routes.

Search results and job mirrors may help discover a candidate, but they can never become a publishable answer on their own.

Quality evidence

The latest clean regression used 100 previously unseen LinkedIn job rows with static resolution memory disabled:

• 98 useful classifications
• 97 safe action routes
• 2 rows safely held as AMBIGUOUS
• 0 unsafe guessed URLs
• 697 total HTTP requests
• 468.6 seconds on the local reference run

The production build also passed private Apify cloud tests:

• 2 normal rows: 2 useful, 2 publishable, 0 held
• 1 unresolved row: 0 publishable, correctly held for review
• 42 automated regression tests passed

These are measured tests, not a promise that every future source will resolve automatically. The product promise is that an unproven route is never marked safe.

Pricing

The Actor charges $2.00 per 1,000 verified job decisions.

• $0.002 per useful decision
• AMBIGUOUS rows are delivered without charge
• failed runs do not create useful-decision charges
• Apify platform usage is included
• no subscription or per-seat fee

Example: verifying 25,000 useful job rows costs $50.

Production workflow

For a recurring feed:

1. Run your existing LinkedIn or job scraper.
2. Pass its dataset ID to this Actor.
3. Publish rows where safeToPublish is true.
4. Send reviewRequired rows to a review queue.
5. Schedule the workflow or connect it through webhooks, Make, Zapier, n8n, or the Apify API.

The default batch is 25 jobs for easy inspection. A run accepts up to 500 jobs. Cold companies take longer because their public website, ATS, and job evidence must be discovered live.

Wire it into Make, n8n, or your own webhook

Add a webhook for the ACTOR.RUN.SUCCEEDED event (Actor page > Integrations). Apify posts this payload to your endpoint the moment a run finishes:

{
  "eventType": "ACTOR.RUN.SUCCEEDED",
  "createdAt": "2026-07-28T15:02:44.906Z",
  "eventData": {
    "actorId": "TW4QDjkomVJH1SPsd",
    "actorRunId": "xYzAbCdEfGhIjKlM1"
  },
  "resource": {
    "id": "xYzAbCdEfGhIjKlM1",
    "status": "SUCCEEDED",
    "startedAt": "2026-07-28T15:00:48.634Z",
    "finishedAt": "2026-07-28T15:02:44.703Z",
    "defaultDatasetId": "qRsTuVwXyZaBcDeF2",
    "stats": { "durationMillis": 115406, "computeUnits": 0.0321 }
  }
}

Your workflow then fetches the verified rows in one call:

GET https://api.apify.com/v2/datasets/{resource.defaultDatasetId}/items?format=json&clean=1

In Make and n8n you can skip the webhook entirely: both have a native Apify integration with a run-finished trigger and a get-dataset-items action, so a scheduled verify-then-publish flow takes two modules.

This Actor is also exposed to AI agents through Apify's MCP server (mcp.apify.com): an agent can discover it by search and run it with the same pay-per-event billing, with no separate integration.

Responsible use and limits

• The Actor checks public web evidence and does not require LinkedIn login credentials.
• Source websites can throttle, remove, or change public pages.
• An active official listing proves that an application route exists. It cannot prove a hiring manager's private intent or guarantee that a role will be filled.
• Review the terms and laws that apply to your data source and use case.