Is this job posting worth your time?
Paste a posting. HireProof scores how closely its wording matches postings labelled as ghost jobs — listings put up with no real intent to hire. Everything runs in this browser; nothing is uploaded.
Try an example:
Hover or tap a marked phrase to see which signal it fed.
Result
Paste a posting on the left and the score appears here as you type.
Resemblance to labelled ghost postings
What moved the score
toward real toward ghost
All eleven signals, ordered by effect. Points is what the score would lose if this posting were merely average on that one signal — so they do not sum to the total.
How it works
Five steps, on text you can edit
The same five stages the scorer runs, on a line you can change.
Model card
What this model is, and what it is not
Every number here is read from model.json. Retraining
updates this page; nothing on it is typed by hand.
What it reads
Wording only. It cannot see how long a posting has been up, whether it has been reposted, or anything about the company — which are the strongest ghost-job signals there are. It is reading the advert, not checking the job.
| Signal | Weight | Direction |
|---|
Where it gets things wrong
- A well-written ghost posting. A listing put up with no intent to hire, but written carefully, reads like any other careful listing. The model has no way to tell them apart.
- A badly-written real job. Plenty of genuine postings are vague and jargon-heavy, especially from small employers without a recruiter. Nothing here distinguishes bad writing from bad faith.
- Short postings. Under about 80 words there is too little text for the signals to mean anything.
- Tech-sector wording. Most labelled postings are software and office roles. Trades, healthcare, hospitality and academia phrase things differently.
- One person's judgement. Every label reflects one reader's call, so the model inherits their blind spots along with their reasoning.
- English only. The phrase lists are English; anything else scores as though the wording were missing.