Ghost Job Warning Signs 2026 — How to spot postings that will never interview anyone

Why ghost jobs are a real problem in 2026

Greenhouse's 2026 update to its State of Job Hunting research reports that three in five candidates suspect they have encountered a ghost job. Greenhouse platform data, drawn from postings hosted on its own ATS, classifies 18-22% of jobs posted in a given quarter as ghost jobs. The Congressional Research Service defines ghost-job postings as listings for positions that do not exist or are not planned to be filled immediately, and the BLS JOLTS definition of a job opening (a specific position, available work, a start window inside 30 days, active recruiting) provides a useful baseline for what a legitimate opening should look like.

None of those sources lets you decide whether one specific posting is fake. They tell you that the market-level frequency is high enough that a serious applicant should screen before tailoring. The mistake is treating every batch of 100 applications as if every posting was equally real and equally worth a custom package.

The four red-flag pattern categories RoleWorth flags

RoleWorth's ghost-risk detector is regex pattern-based, not machine learning. It reads posting text and flags four specific categories. Pattern detection has known limitations (false positives on legitimate roles using unfortunate language, false negatives on ghost roles written by careful recruiters) which is why these signals feed into the Worth Score as one factor rather than producing a final verdict.

1. Commission-only language

Phrases like "100% commission," "unlimited earning potential," "eat what you kill," "1099 contractor only," and "no base salary" correlate with two bad outcomes: roles that are structurally tilted toward MLM-style recruiting, and roles where the employer has no headcount cost and therefore no incentive to close the loop with candidates. Commission-only is legal and sometimes appropriate (high-end real estate, certain sales roles) but the language pattern is also a reliable indicator of evergreen pipeline collection. Treat commission-only postings as Maybe unless you can verify the team, quota, territory, average rep tenure, and ramp guarantee through a direct conversation.

2. Training-fee and pay-to-work scams

Any posting that asks the candidate to pay for training, certification, equipment, background-check fees, or onboarding materials before starting work triggers this category. The pattern often hides behind language like "onboarding investment," "starter kit," "certification required (we provide at cost)," or "refundable after first paycheck." The US FTC has documented job-scam patterns where the entire revenue model is the training fee, not the work product. This is the highest-severity category in the four — a hit here should usually mean skip, not maybe.

3. Vague-scope filler

This category catches postings whose responsibilities section is filler: "wear many hats," "fast-paced environment," "rockstar / ninja / guru," "help us build the future," "cross-functional collaboration," with no concrete outcomes, scope, tools, team, or product surface attached. Vague scope correlates with two failure modes: postings copied from a generic template for talent-pool collection, and postings written by hiring teams that have not yet aligned on what the role is. Either way, your custom resume cannot make the match obvious to a reader who has not been told what the role is. Treat vague-scope postings as Maybe until you can pull at least three concrete responsibilities from the recruiter, manager, or company source.

4. Low-specificity urgency

Manufactured urgency without specifics: "urgent hire," "immediate start," "need to fill ASAP," "multiple positions available," "limited spots" — combined with no posting date, no hiring-manager name, no team, no start month, and often no compensation band. Real urgency in a real hiring process is usually attached to a specific event (a launch, a quarter close, a parental leave backfill, a contract start). Generic urgency without anchors correlates strongly with talent-pool reposts and aggregator-only listings.

Seven additional warning signs from public research

The four pattern categories above are RoleWorth's automated check. Below are seven additional signs that require a manual read of the posting and a quick check against the company's own site. None is a verdict on its own. Three or more together should change your next action from "tailor" to "ask one clarifying question first."

5. Stale repost dates

A posting older than 30 days with no visible update trail is a yellow flag. A posting that has been reposted three or more times in 90 days, with the same title and no clear team change, is a red flag. The repost pattern is how aggregators and pipeline-collection roles stay visible without committing to interviewing. Some ATSes show a true "originally posted" date; many job boards show only the most recent crawl date, which can hide the staleness.

6. No hiring-manager name or team

Real openings usually have an owner. The owner's name does not have to be public, but the team, function, manager level, reporting line, product area, or segment should appear somewhere in the posting. If the role says "join our growing team" without naming the team, the role is either deliberately generic (talent pool) or written before the hiring decision was finalized. Either way, your tailored resume cannot prove fit because the requirements are not stable.

7. No clear hiring timeline

BLS JOLTS classifies an opening as a position where work could start within 30 days. If the recruiter cannot answer "when are you scheduling first-round interviews?" with a window inside the current month, the posting may be live but not currently moving. Postings on indefinite hold are not lies — but they are not active openings either. A short clarifying question ("Is this role actively interviewing this month?") saves several hours of tailoring effort.

8. Missing or absurdly wide salary band

Salary transparency laws now cover postings to candidates in California, Colorado, Hawaii, Illinois, Maryland, Minnesota, New York, Washington, and several other US jurisdictions, with similar disclosure rules in the EU pay-transparency directive scheduled for member-state transposition by 2026. A posting that lists no band when one is legally required is a compliance risk for the employer, which means the posting may not have been reviewed recently. A band so wide it covers two levels (for example "USD 90,000–220,000") usually means the role is open at multiple levels or the band was copy-pasted from a different role.

9. Applications routed through talent-pool boards

"Join our talent community" and "general application" pages are explicit talent-pool collection. They are honest about their purpose, which is helpful — but the time-to-interview from a talent pool is structurally longer and the application itself is rarely tied to an open requisition. Treat talent-pool submissions as low-effort networking, not as a real application.

10. Company source does not match the job-board source

Open the company's own careers page (greenhouse-hosted at boards.greenhouse.io/[company], Lever at jobs.lever.co/[company], Ashby at jobs.ashbyhq.com/[company], or the company's direct /careers URL). Search for the role title. If the role does not appear, or appears with a different title, location, level, or set of responsibilities, the job-board copy is suspect. Trust the canonical company-source URL over the aggregator.

11. Dead, looping, or mismatched application links

The application button leads to a dead error page, redirects through several aggregators, lands on a different role, or asks you to re-enter information already in your resume on a domain that does not match the company. A broken application path almost always means the role is no longer being maintained. This is one of the strongest single signals — an employer that is actively interviewing fixes broken application links within days.

How to score the warning signs

Low risk (0-2 weak signals): the role is fresh, the company source confirms it, responsibilities are concrete, logistics are clear, and you can map at least three proof points to must-have requirements. Apply with a tailored package if the role is otherwise worth it.

Medium risk (3-5 weak signals):the company may be real and the role may exist, but the posting is stale, thin, or logistically unclear. Send one precise clarifying message before spending a high-effort application unit. Sample message: "Is this role actively interviewing this month, and is the remote scope US-only or global?"

High risk (6+ weak signals, or any scam-adjacent hit): training-fee language, suspicious contact paths, off-platform payment requests, unusual personal-data collection, or mismatched domains. Skip or keep it as a low-priority watch item unless you can verify the role through an official company channel.

What RoleWorth actually does — and does not do

RoleWorth's Ghost Job Detector reads the posting text and runs the four regex pattern categories above. It surfaces matched patterns in the Worth Score breakdown so you can see whichpatterns triggered, not just a single risk number. It does not claim to know an employer's internal intent. It does not promise that flagged jobs are fake. It does not promise that unflagged jobs are real. It treats ghost-risk as one input to the Apply / Maybe / Skip decision.

What it does not do: it does not use machine learning, it does not call a third-party intent-scoring API, and it does not maintain a private list of "known ghost employers." The detection is transparent — the patterns are visible in the score breakdown, and you can override the recommendation with your own judgment.