Ghost Jobs by the Numbers: 29.7% of Tech Postings Sit Open Past 60 Days

Everyone in tech has the same suspicion: a lot of the jobs advertised online are not really being filled.

The suspicion is hard to prove. You cannot see a hiring manager's intent from a job ad. What you can see is how long the ad stays up.

So we measured that. Every day since 20 June 2026 we have re-checked every posting in our index and recorded the day it disappears. As of 4 September 2026 that is 19,830 postings across 1,680 companies, each verified daily for up to 76 days.

Here is what the data says.

Nearly a third of postings are still open after two months

Of the 19,830 postings we track, 5,893 were still publicly open more than 60 days after we first saw them. That is 29.7%.

Widen it slightly and the picture holds:

This is not a fringe phenomenon confined to a few bad actors. 1,019 of the 1,680 companies we track, roughly 61%, have at least one posting that has been open past 60 days.

A posting open for two months is not proof of a ghost job. Senior and specialised searches genuinely take that long. But it does mean something concrete for you as a candidate: that listing has already failed to close once, and you are applying into two months of accumulated competition.

Remote roles close much faster than onsite ones

The sharpest split in the data is not seniority or salary. It is location.

Onsite and hybrid roles go stale about a third more often than remote ones.

We are not claiming to know why. The plausible explanations point in different directions: remote roles draw from a national rather than local pool and therefore fill faster; or onsite listings are more often used to keep a local pipeline warm. Either way, if you are choosing where to spend your applications, the remote listing is measurably more likely to be a real, closing search.

A salary range does not mean the job is real

Pay transparency laws have pushed disclosure up. In our data 60.2% of postings publish a salary range, with a median band of 150,000 to 190,000 US dollars.

That is good for candidates. It is not, however, a signal of seriousness:

Postings with a published range go stale slightly more often. The difference is small and we would not lean on it, but it firmly kills the intuition that a company willing to name a number must be ready to hire.

The number we are not going to publish

You will read a lot of confident claims about how long the average job posting lasts. Numbers like "the average posting is filled in 12 days" circulate widely.

We could have published one. We are not going to, because we think the standard way of calculating it is wrong, and ours would have been wrong too.

Here is the problem. Only 46.7% of the postings we track have ever closed. The other 53.3% are still open, and their median age is already 66.8 days. If you compute an average lifespan from the closed postings alone, you are throwing away every long-lived listing that has not finished yet, and those are exactly the ones that would pull the number up.

It is like measuring how long people stay at a party by surveying the ones who already went home.

The effect is not subtle. 8,756 of the postings that are still open are already older than the "median" you would get from the closed ones. Any average built that way is guaranteed to be too low, and the more ghost jobs exist, the more badly it understates them.

Measuring this properly requires survival analysis, which handles the still-open cases rather than discarding them, and a tracking window long enough that most postings have resolved. We have 76 days. We will publish a duration figure when we can defend it.

We are flagging this because the figure we almost published was our own, and someone else's version of it is probably in an article you have read.

What this means if you are job hunting

• Check the posting date before you invest an hour in the application. Past 60 days you are in the stale third.
• Weight remote listings up. They close a third more often.
• Do not read a salary range as a sign of urgency. It is a legal requirement in many places, not a signal.
• Speed beats polish at the front of the distribution. Roughly a third of the postings that close are gone inside a week.
• Look at the company, not the posting. Six in ten companies are sitting on at least one stale listing. Several is a pattern.

Method, and what it cannot tell you

We fetch postings from company career pages and applicant tracking systems, then re-verify each one daily and record the date it stops appearing. Daily re-verification is what makes duration measurable at all; a one-time scrape cannot produce any of the numbers above.

Three limitations we want to state plainly:

1. Disappearance is not the same as filled. A posting can come down because the role was cancelled, frozen, or reposted under a new ID. We measure visibility, not outcomes.
2. The window is 76 days. Every figure here is bounded by that. Long searches are under-represented by construction.
3. Our index is not the whole market. It skews toward software and adjacent technical roles at companies that publish structured listings. It is not a census.

Our seniority labels are also too coarse to break out reliably, so we have deliberately left that cut out rather than present a split we do not trust.

Aggregate data available to journalists and researchers on request.

Preparing for the interviews that are real

Filtering out stale listings is only the first half. The applications that do convert still end in a live interview, usually a harder one than the same role demanded two years ago.

That is the part InterviewChamp.AI was built for: real-time help during the actual conversation, in your own words, so the interview you finally earn is one you are ready for.