The February 2026 JOLTS release put 6.9 million openings on the board and 4.8 million hires against them. That is a 2.1 million per month gap between what is advertised and what actually converts, and it is why your inbox has gone quiet even though the headline numbers still look busy.
The strategic response for a job seeker in 2026 is not more applications. It is throwing out postings-as-signal and re-anchoring on reqs you can verify are alive.
What the 2.1M phantom gap actually is
The phantom gap is the monthly delta between BLS-reported openings (6.9M in February 2026) and BLS-reported hires (4.8M). Hires fell 498,000 month-over-month and 387,000 year-over-year, dropping the hires rate to 3.1%, its lowest reading since April 2020.
Two mechanics drive it:
- Openings are counted if the employer says the role is recruitable within 30 days. No hire has to happen. No interview has to happen. A req that gets budget-frozen on day 29 still counts.
- The quits rate has been at or below 2.0% for eight straight months and hit 1.9% in February. Almost nobody is voluntarily leaving, which means fewer backfill reqs (the kind that almost always close) and more net-new headcount reqs (the kind that get paused).
6.9M openings against 4.8M hires, with the hires rate at its lowest since April 2020.
Indeed Hiring Lab called the February report "stuck in neutral." That is generous. For applicants it is worse than neutral, because the openings number is what recruiters, career coaches, and LinkedIn feeds still cite when they tell you the market is fine.
Why "ghost jobs" only explain a quarter of the collapse
Ghost jobs are real, but they are quantitatively small compared to the story people are telling about them. Revelio Labs found hires per posting fell from roughly 0.80 in 2019 to 0.40 in 2024. If you assume one posting in five is a pure ghost and eliminate every one of them, the yield lifts from 0.40 to only about 0.50. That recovers roughly 25% of the decline.
The other ~75% is real reqs where the employer genuinely tried and no hire happened. Mechanisms:
- Budget freeze mid-cycle
- Req reorg or hiring-manager change
- JD over-specced past what the market can supply
- Recruiter throughput ran out before a decision got made
- Offer extended, candidate said no, req quietly shelved
Ashby's ATS data across 22,000+ reqs from 2021 to 2024 lines this up. In 2024, 82% of open jobs were filled and 18% were not. Of the unfilled 18%, only 5.5% were paused, 1% had an offer made with no hire, and 3.5% had no reason given.
Optimizing purely against "is this a ghost?" solves a quarter of the problem. The other three-quarters is real intent that stalled.
Daniel Grimm, former CFTC senior counsel, built a November 2025 Columbia Law Review Forum piece around the Revelio collapse, arguing ghost postings already violate Section 5 of the FTC Act. The legal argument is fair. The practical takeaway is that ghost detection is table stakes, not strategy. You still need to filter for reqs that will actually convert.
The numbers that should drive your 2026 shortlist
Here is the dataset every 2026 job seeker should be working from, pulled from BLS, Revelio Labs, Ashby, Juvo Jobs, Intervu HQ, and Refolk's own index of professional profiles.
| Metric | Figure | Source |
|---|---|---|
| Openings vs. hires gap, Feb 2026 | 6.9M − 4.8M = 2.1M/month | BLS JOLTS |
| Hires per posting, 2019 to 2024 | 0.80 to 0.40 | Revelio Labs / Columbia Law Review |
| Ghost-only recovery ceiling | 0.40 to 0.50 | Derived, Business Model Analyst |
| Share of yield collapse not from ghosts | ~75% | Derived |
| Ashby fill rate, enterprise vs. all | 89% vs. 82% | Ashby ATS report |
| Senior SWE to Technical Recruiter ratio, US | 8.8 to 1 | Refolk index, Aug 2026 |
| Application response rate on Indeed | 20 to 25% | Juvo Jobs / Fast Company |
| Quality lift when salary is disclosed | +24.5% across dimensions | Intervu HQ, 2026 |
Two of these are the ones nobody is quoting yet. The 89%-vs-82% Ashby split flips the standard "apply to startups, they move faster" advice on its head for a low-hire market. And the 8.8-to-1 senior-engineer-per-recruiter ratio explains, arithmetically, why "quality over quantity" is not soft advice.
How to spot fake job postings in 2026
The best ghost filter is not posting age. It is the combination of salary disclosure, ATS source, quits-rate context, and language that implies a seat is empty. Run every posting on your shortlist through these five signals before you spend an hour tailoring for it.
- Salary band disclosed. Postings with comp bands score 24.5% better on every quality dimension Intervu HQ tracks. Comp bands require finance sign-off, and finance sign-off means an approved req.
- Applied via the raw ATS URL, not a LinkedIn or Indeed repost. Ashby, Greenhouse, Lever, and Workday career pages are the source of truth. If the LinkedIn "apply" button routes you to a job that no longer exists on the company's ATS, that posting is dead. Workday and Taleo also auto-renew every 30 to 90 days, so a January posting can look fresh in March even if it closed in February.
- Backfill language, not growth language. "Replacing," "expanding an existing team," and "reports to [named person]" imply a real seat. "Building out a new function" and "founding member" imply headcount that has not been approved yet.
- Enterprise employer, especially on Ashby. 89% of enterprise reqs on Ashby filled in 2024. 97.5% of jobs at companies with 50+ employees moved at least some candidates to active interview stage. Under 50 employees, the odds get worse fast.
- Not in a sector with a known ghost-rate spike. Government roles run ~60% ghost rate per the Columbia Law Review analysis. Information and tech run ~48%. If you are applying to a federal or tech req with none of the four signals above, treat it as pipeline-building.
Clarify Capital's January 2025 survey of 1,000 employers found nearly 1 in 3 admit posting jobs with no current hiring intent. A separate figure cited in the Columbia Law Review piece: 40% of hiring managers said their company had run ghost listings in the past year, and 30% were maintaining active fake listings at the time they were surveyed. The filters above exist because the admissions are that blunt.
The recruiter-throughput ceiling nobody wants to admit
There is a hard arithmetic limit on how many applications ever get read, and it has nothing to do with your resume. Refolk's index of US professional profiles shows roughly 181,000 people currently titled "Senior Software Engineer" and only about 20,600 titled "Technical Recruiter." That is 8.8 senior engineers per technical recruiter, before you count staff, principal, or manager tracks.
From Refolk's index of ~181,000 senior engineers against ~20,600 technical recruiters.
The consequence: on a req that draws 300 applicants, a single recruiter has maybe an hour of screening time budgeted. They will read the top 5 to 10% surfaced by the ATS ranker, plus whatever came in through referral. Your resume is not competing with the pool. It is competing for a slot in the recruiter's top 15 to 30.
That is the exact friction Refolk is built for: paste the posting, get your own resume back rewritten so the ATS ranker actually surfaces you for that req instead of the generic version you sent to the last 40. The alternative is spraying 300 applications at postings that were never going to hire anyone, at a 20 to 25% response rate on the ones that were.
A shortlist strategy that beats spraying 300 applications
The 2026 move is a 20-role shortlist rebuilt weekly, not a 300-role blast. Here is the structure that works for a market where the openings-to-hires ratio is 1.44 and the quits rate is 1.9%.
Step 1: Build the shortlist from ATS pages, not aggregators
Go directly to boards.greenhouse.io, jobs.ashbyhq.com, jobs.lever.co, and the target company's Workday tenant. Filter for the five signals above. Twenty roles per week is the ceiling for anything you can meaningfully tailor.
Step 2: Tailor the resume per posting, not per company
The Ashby data is clear: 97.5% of 50+ employee reqs move someone to an active stage. If you get filtered out at ATS scoring, you never enter the pool the recruiter actually reads. Rewriting each resume by hand for 20 postings a week is unrealistic, which is where Refolk earns its slot in the workflow. Paste the posting, get a resume tailored to it, get a cover letter drafted against the same JD, and get a fit score that tells you whether the posting is worth the send in the first place.
Step 3: For enterprise reqs, add a targeted intro
Enterprise fills at 89%. If the req is real and you can get a warm intro to the hiring manager or a peer engineer on the team, you skip the recruiter throughput bottleneck entirely. Two intros per week beats 30 cold applies.
Step 4: Drop postings drifting toward the unfilled bucket
Ashby's data implies most enterprise reqs either move to interview stage within a month or drift toward the 18% unfilled bucket. If a posting has been up more than 30 days on the raw ATS URL and you cannot find a single LinkedIn post from anyone at the company about the role, drop it from the shortlist.
The advocacy layer is catching up. New Jersey Assembly Bill 4625 and California AB 1251 (2025 to 2026 session) would both require employers to disclose whether a posting is for an existing vacancy and force removal once filled. Truth in Job Ads (truthinjobads.org) is tracking the state-level fights. Until those pass, the signals above are the closest thing you get to a disclosure requirement.
FAQ
Are ghost jobs illegal in 2026?
Not federally, but the argument is getting louder. Daniel Grimm's November 2025 Columbia Law Review Forum piece argues ghost postings already violate Section 5 of the FTC Act as deceptive practices, and New Jersey Assembly Bill 4625 and California AB 1251 in the 2025 to 2026 session would require employers to disclose whether a posting is for an existing vacancy. Until one of those passes and gets enforced, the burden is still on the applicant to filter.
What is a realistic response rate on job applications in 2026?
Juvo Jobs research published in Fast Company puts the response rate on major job boards like Indeed at 20 to 25% of applications. That is any response, including automated rejections. The rate is meaningfully higher on raw ATS applications to enterprise employers with salary bands disclosed, and meaningfully lower on aggregator reposts of postings older than 30 days. If you are getting under 10%, the problem is almost always the posting mix, not the resume.
Should I still apply to postings without a salary band?
Sometimes, but weight them lower on your shortlist. Intervu HQ's 2026 corpus shows postings with salary information score 24.5% better across every quality dimension they track, and the mechanism is simple: comp bands require finance sign-off, which requires an approved req. A no-band posting can still be real, especially at pre-IPO startups in states without pay-transparency laws, but it is a weaker signal than a banded posting from the same stage.
How many applications should I actually send per week?
Twenty tailored applications to reqs that pass the five signals, plus two warm intros, beats 300 cold applies. If ghost jobs plus stalled reqs mean roughly 60% of postings never hire anyone, and recruiter throughput caps the readable pool at the top 5 to 10%, then only a handful of your 300 blast applications were ever going to be read by a human. Twenty carefully filtered applies puts you in that same readable pool with an order of magnitude less work.