◆TDW

Why the Job Search Is Broken, and What Would Fix It

Fewer openings, ghosting, ghost jobs, hidden salaries and screening on the wrong signals. What the data says about each problem, which fixes have evidence behind them, and which ones a job board can actually build in.

Ask anyone who has looked for a tech job since 2023 and you get the same list: hundreds of applications, few replies, listings that turn out not to exist, salaries revealed only at the offer stage. Some of that is the market. Some of it is how hiring is set up, and that part can be changed. This post goes through each problem with the data behind it, then looks at which fixes have evidence and which are still guesses.

1. There are fewer openings, and junior roles took the biggest hit

Start with the part nobody can design away. On Indeed, US tech job postings more than doubled between 2020 and early 2022, then fell to 36% below their February 2020 level by July 2025 (Indeed Tech Talent Report 2025). Software development postings bottomed around 34% below the 2020 baseline in early 2025 and have since partly recovered, to about 20% below in October 2026 (FRED, Indeed software development postings index). Other tech categories were weaker for longer: in January 2026, data and analytics postings were still 40% below 2020 and falling (Indeed Hiring Lab, Q4 2025).

The drop wasn't spread evenly. In February 2025, tech postings with junior or standard titles were 34% below pre-pandemic levels, against 19% for senior and manager titles (Indeed Tech Talent Report 2025).

These are postings, not hires, and the figures are US-only. They still explain the baseline: more people chasing fewer ads, with the squeeze worst at the entry level. Everything below makes that harder than it needs to be.

2. Applications disappear without an answer

In Greenhouse's December 2024 survey of 2,500 workers in the US, UK and Germany, 61% of job seekers said they had been ghosted after a job interview, up nine percentage points since April 2024. Not after an application, after an interview (Greenhouse, 2024 State of Job Hunting).

The same report shows the feedback loop behind it. Greenhouse says recruiter workload on its platform rose 26% in a single quarter, and 38% of job seekers said they were mass-applying to roles. Easy-apply buttons and AI-written applications make each application nearly free to send. Recruiters respond to the volume by answering fewer people, and candidates respond to the silence by sending more applications.

The underlying cause is that ignoring an applicant costs the employer nothing. No standard says how long a company has to answer, and the candidate can't see which companies answer and which don't. Our stage-by-stage timeline of a remote job search covers what normal delays look like.

3. Some listings were never meant to be filled

A ghost job is, in the US Congressional Research Service's words, an online posting "for positions that do not exist, or that employers are not planning to fill immediately." CRS notes that there are no official statistics on how common they are, and that the industry estimates are hard to assess because methods vary and the firms publishing them sell hiring services (CRS, IF12977).

With that caveat, the numbers people cite are these. In a May 2024 Resume Builder survey, 40% of 1,641 hiring managers said their company had posted a fake job in the past year. That is a share of companies, not of postings (Bloomberg Law). Greenhouse says that in any quarter, 18–22% of the jobs posted on its platform are classified as ghost jobs, and three in five candidates in its survey suspected they had run into one (Greenhouse).

Lawmakers have started to act. New York bill S8877 would require job ads to say whether the vacancy is real and when the employer expects to hire, and filled ads would have to come down within two weeks. It passed the Senate in April 2026 and the Assembly in June 2026; at the time of writing it had not been signed. Similar bills are pending in New Jersey, Kentucky, California and Pennsylvania. Ontario's vacancy-disclosure and applicant follow-up rules took effect on January 1, 2026 (Bloomberg Law; CRS). Our post on verifying a remote listing, a recruiter and an offer covers the related problem of listings that are outright fraud.

4. Salaries are hidden until it's too late

When a posting gives no salary, you can go through three interview rounds before learning the role pays less than you earn now. This one has a tested fix. Colorado was the first US state to require salary ranges in job postings, from January 2021. A study using job-posting data found the share of Colorado postings with salary information went from about 35% before the law to nearly 70% a year later, while other states stayed at 30–40%. Among employers that posted salaries both before and after, posted pay rose about 3.6% on average, and ranges didn't get wider to hide the real figure (Arnold, Quach and Taska, 2023). Compliance was incomplete, and other studies of the same law estimate a different size of effect, but the direction is consistent.

Europe is following. Under the EU Pay Transparency Directive, applicants are entitled to the initial pay or its range for a position before the interview, and employers may not ask applicants about their pay history. Member states had to bring the directive into national law by 7 June 2026 (Directive (EU) 2023/970, Articles 5 and 34).

5. Candidates are screened on the wrong signals

The first screen often looks at things that have nothing to do with whether you can do the job.

Names. In one of the largest hiring experiments ever run, Kline, Rose and Walters sent more than 83,000 fictitious applications to 108 of the largest US employers. Applications with distinctively Black names were 2.1 percentage points less likely to get a reply. The effect was concentrated: the fifth of firms that discriminated most accounted for nearly half of the lost contacts, and firms with centralised hiring discriminated less (Kline, Rose and Walters). The jobs were mostly entry-level retail and service roles, not tech, but the mechanism, a first screen that sees the name before the skills, is the same.

Exact matches. In a Harvard Business School and Accenture survey of 2,275 executives, 88% agreed that qualified high-skills candidates are screened out because their applications don't exactly match the criteria in the job description (HBS Working Knowledge). Our post on why most resumes get filtered out covers how that filtering works.

Anonymous CVs are not a simple fix. In a German pilot, anonymised applications gave women and applicants with a migration background equal interview chances, and helped where they had been disadvantaged before (IZA Research Report 44). A French randomised trial found the opposite: the gap between minority and non-minority candidates widened, because the firms that volunteered had already been favouring minority candidates, and anonymity removed that (J-PAL summary of Behaghel, Crépon and Le Barbanchon). Hiding the name helps when the screen is biased against the candidate. It also hides any context an employer was using in the candidate's favour.

6. "Remote" doesn't always mean remote

The research found no reliable dataset on how often "remote" postings turn out to be hybrid or limited to certain countries, so there's no number here. The pattern itself is familiar: a role labelled remote that requires residence in one country, office days, or working hours that only fit one time zone. The cause is that "remote" is one word covering several different arrangements, and the restrictions sit in the description text, if they are written down at all. Our post on why remote jobs turn out hybrid or region-locked goes through the signals to look for.

What would fix it

Each problem above has a cost that falls on the candidate and none that falls on the employer. The fixes that have evidence all do the same thing: make the employer state something up front, or make silence visible.

ProblemFixEvidence
GhostingA response deadline, and public response ratesNo study yet. Ontario now requires applicant follow-up, so data should follow
Ghost jobsVacancy disclosure, removal of filled ads, a cost to keep a listing upLaws in force in Ontario, pending in several US states. No outcome data yet
Hidden salariesMandatory salary rangesColorado: salary disclosure roughly doubled, posted pay up about 3.6%
Name biasAssess skills before identity is shownStrong evidence of the bias; mixed evidence on anonymity as the fix
Exact-match filteringStructured requirements instead of keyword matchingEmployers themselves report the problem (HBS); no controlled study of the fix
Vague remote labelsWork type and region as fixed fields, not proseNo data either way

The market problem in section 1 has no fix on this list. A job board can't create openings.

What The Dream Work does about it

We built the platform around the fixes in that table. To be specific about what it does:

  • A five-day answer window on both sides. A company has five days to answer an application. If it doesn't, the application closes automatically and the candidate is told. When a company asks to connect, the candidate has five days to answer. Every company's public profile shows its response rate and average reply time, so you can see who answers before you apply. How applications move after you apply has the details.
  • A salary range on every listing. Companies can't publish a listing without one.
  • Skills before identity. The company sees a candidate number, the job match score and its breakdown. Your name and email are shared only when both sides agree to connect.
  • Structured requirements, no CV. Companies mark which skills are required and which are nice to have. You're matched on a structured profile, and every point of the score is explained, so there's no keyword filter to get past.
  • Work type and region as fields. Remote, hybrid or on-site and the allowed region are set per listing and count toward the score, so a region-locked role shows up as such before you apply. The formula shows how much each one counts.
  • Listings cost money and expire. Each listing runs for 30 days and costs the company a flat fee (the first one is free), so keeping an evergreen listing up costs something. Candidates don't pay anything.

What it doesn't fix:

  • The number of openings. The platform is new, and it has far fewer listings than the large boards.
  • Everything after you connect. Interviews, take-home tasks and offers happen off the platform, and the five-day rule stops applying there.
  • Bias after identity is shared. Anonymity covers the first screen. As the French trial shows, it isn't a complete answer, and the platform doesn't claim it is.

If you want to see the approach in practice, browse the remote tech jobs that match your skills or the full list of tech jobs. For the wider question of where to look, start with where and how to find the best remote tech jobs in 2026.

Limits of the evidence

The posting data is US-only and counts job ads, not hires. The ghosting and ghost-job figures come from surveys run by companies that sell hiring software, and none are official statistics. The bias and anonymity studies are not specific to tech, and the anonymity trials are more than ten years old. The legal status of the US ghost-job bills changes quickly; check the bills' own pages for the current state. No verified data was found on how often employers ghost after an application (as opposed to after an interview), on AI-written applications' effect on reply rates, or on how often remote labels are misleading.

Sources

For candidates

See your match score on every open role.

Build your profile in 5 minutes and get scored against every active role.

Create free account →