Why your best candidates never apply to job posts

By SkavanderHR · 13 September 2026

A job post reaches people who are actively looking. Most people who could do a hard to fill job are not looking, and developer surveys put that share above four in five in the UK, Germany and the US. A quiet post is rarely a writing problem. It is pointed at the wrong crowd.

Key takeaways

• Stack Overflow’s own developer surveys have repeatedly found most developers are not actively job hunting: above four in five in the UK, Germany, the US and Canada in the 2022 wave, with at least half of those still open to the right offer.
• Volume has risen, not quality. Greenhouse’s own platform data averages 254 applicants per posting.
• In Germany, a third of companies with open IT roles say applicants do not meet the technology requirements, and almost as many get few or no applications at all.
• Peer reviewed research finds individual performance follows a power law rather than a bell curve, so the few strongest people in a market matter more than an average CV suggests.
• Reaching people who are not looking is a search task, and it looks nothing like writing a better advert.

Who is actually reading your job post?

The readers of any job post are the people who went looking for one that week. That group is real, and for a common role with a deep applicant market it is enough. It skews, though, towards people between jobs, people early in their careers, and people unhappy enough where they are to spend an evening on a job board.

The people who are not in that group are the majority. Stack Overflow has asked developers about this for years: in the 2022 wave, more than four in five respondents in the UK, Germany, the US and Canada described themselves as not actively looking, while at least half of them said they remained open to a new job. That second half is the part worth holding on to. Not looking is not the same as not interested. It means nobody is checking job boards, so a post is simply never in front of them.

Two honest caveats belong here. This is a self reported survey of one profession, so read it as the shape of a market rather than a precise count. And the evidence says nothing about anyone’s quality: there are excellent people actively looking, and being employed is not a proxy for being good. The defensible claim is narrower and it is about reach. Most skilled people who could do the job are not watching your careers page.

If more people apply, why is the shortlist worse?

Because the two are not connected. Greenhouse’s own platform data for 2025 puts the average posting at 254 applicants, with applications per recruiter up 412% over the period it measured. More arrivals per post, from a pool that did not change, means a lower share of them fit.

The German picture makes the same point from the employer’s side. Bitkom Research, the German tech industry association, surveyed companies in the first half of 2026: 97% of those with open IT roles report difficulty filling them, 33% say applicants do not meet current technology requirements, and 31% get few or no applications for an open IT role at all. Those last two numbers describe two different failures of the same channel. One posting drowns and another gets nothing, and neither has anything to do with the wording.

Volume up, fit flat: what the published figures show

What was measured	Figure	Published by
Applicants per job posting, platform average	254	Greenhouse, 2025
Rise in applications per recruiter	412%	Greenhouse, 2025
German companies with open IT roles reporting difficulty filling them	97%	Bitkom, 2026
Say applicants do not meet current technology requirements	33%	Bitkom, 2026
Get few or no applications for an open IT role	31%	Bitkom, 2026
UK IT employers who struggled to find the skills they needed	75%	ManpowerGroup, 2025

Sources: Greenhouse, 2025 AI in Hiring Report. Bitkom Research press release of the third of September two thousand twenty six, on surveys of 853 and 602 companies. ManpowerGroup UK Talent Shortage Survey 2025, as published by Experis.

The reading time that volume costs is harder to pin down. One vendor synthesis of several sources puts it at roughly 8% of applicants passing the first CV screen, about one hire per 200 applications, and on the order of 23 hours of screening per hire. Treat those as the shape of the problem rather than a measured benchmark, because no single study produced them. The direction is not in doubt for anyone who has read a full inbox.

How much does missing the strongest few people cost?

More than an average would imply, and this is the best evidenced claim on this page. O’Boyle and Aguinis, publishing in Personnel Psychology in 2012, looked at individual performance across 198 samples and more than six hundred thousand people, and found it does not follow a bell curve. It follows a power law: a small number of people produce a disproportionate share of the output, consistently across industries, job types and measures.

What that does not give you is a multiplier. McKinsey summarises the same body of research as high performers being up to eight times more productive in complex roles such as software development, and that framing is McKinsey’s, not a measurement in the original paper. Either way the practical point survives. In a market of a few hundred qualified people, the difference between the top handful and the median is not a rounding error, and a channel that reaches only the ones who happened to be looking this month will miss most of the handful by construction.

What does this mean for a company of twenty to five hundred people?

Three things, and none of them require abandoning the post you already have.

First, separate the two failure modes before you act. Applications arriving with nobody worth interviewing is a fit problem, and the fix sits in the brief and the screen. No applications arriving at all is a reach problem, and no rewrite touches it.

Second, size the market rather than the inbox. For a narrow role, there is a countable number of people in your geography who genuinely fit: often a few hundred, sometimes fewer. That number, not the applicant count, is the real denominator of the search.

Third, be honest about who is going to do the reaching. A lean team without a dedicated recruiter has no spare hours for a person by person search, which is exactly why the default is to post and wait, and why the wait keeps not working.

What changes if you go and find them instead?

We run the other side of that. The system searches the professional networks and company websites for every person in the market who matches the criteria for one open role, and each of them is told about the role about that one role. Nobody has to be looking for a job for this to work, which is the whole point: it reaches the four in five who are not. The people who answer are engaged before your team spends an hour on them, typically within two weeks of starting, and it costs about a tenth of a recruiting agency. Your own team still takes every interview.

If the choice is still open, it is worth reading what an agency, an RPO provider, an embedded recruiter and a monthly search each actually do before picking one, and how the search runs in practice. For a role such as data platform engineering, where the people who fit are certified on one specific product and almost never on a job board, the post was never going to find them. Who it reaches was always the constraint.

Numbers from our own searches. Written and checked by SkavanderHR.