Trender & Insikter
Ta del av våra insikter och trender inom mångfald, rekrytering och ledarskap.

Visa alla inlägg
16 oktober 2026

Talent isn't the problem. Selection is.

New research: 3,048 job applications reveal where talent gets filtered out.

We often talk about skills shortages. But new research shows that much of the talent isn't missing at all – it gets filtered out before anyone is even invited to an interview. And neither more training for the candidate nor a written policy from the employer is enough to change that.

Same qualifications. Different answers.

In the first half of 2025, economist Nick Drydakis sent 3,048 realistic applications to 1,016 real job openings in London. Each job received three applications with the same education, experience and quality. The only difference was one detail about the person: age, ethnic background, sexual orientation, or disclosing that they are autistic.

The result was clear. Candidates who differed from the norm were on average 13.7 percentage points less likely to be invited to an interview – with exactly the same qualifications on paper. The pattern appeared in all four groups.

More skills only helped partly

A common response to gaps like this is that candidates should upskill. The researchers tested exactly that. The same candidates were given a six-month, job-relevant AI certificate on their CV.

It helped, but only partly. The gap narrowed from 13.7 to 9.7 percentage points. Roughly 29 percent of the disadvantage disappeared – the rest remained. For one of the groups, the study could not detect any effect at all.

Another recent study points in the same direction. In a randomised trial in Colombia, job seekers had their occupational skills formally certified. Employment rose by 5.7 percentage points among locals, while no effect could be detected for people whose experience came from another country. A credential has value – but on its own, it doesn't decide how a candidate is assessed.

The disadvantage follows all the way

The gap didn't stop at the interview. Candidates who differed from the norm were also more often invited to lower-paid positions. The certificate narrowed that gap too, but didn't close it.

A policy is not a process

Around 31 percent of the employers in the study had a written equality and inclusion policy. Yet there was no measurable difference in who they invited to interview compared with employers without one. The researcher is careful to note that this doesn't make policies meaningless. But a policy describes an intention. It is the method of selection that decides the outcome.

What it means for you as an employer

The conclusion is both uncomfortable and hopeful. Uncomfortable, because talent is lost in every selection process that relies on gut feeling. Hopeful, because it can be fixed – and the solution doesn't require more candidates, just a better assessment of those who already apply.

  • Criteria before CVs. Define what the role actually requires before reading the first application.
  • The same assessment for everyone. Structured questions and work samples, scored against the same requirements.
  • Measure who moves forward. Track every step of the selection, not just who was hired in the end.
  • What we see every day

    It is also what we see in our own work. The best candidates are rarely the ones who stand out first – they are the ones who get a fair chance to show what they can do. That is why we work from our framework, The Social Sourcing Engine™: we start with the requirements, search widely even where others don't, and assess everyone against the same criteria. It has delivered more than 10,000 recruitments, creating value for our clients and for the communities around them.

    Sources: Drydakis, N. (2026). Artificial Intelligence-Related Digital Skills and Employment Outcomes for Underrepresented Women. IZA Discussion Paper 18964. Preliminary research paper, conducted in London in 2025 with female applicants to full-time private-sector roles not requiring a university degree. Busso et al. (2026). Skill Certifications and Migrant Labor Market Integration: Experimental Evidence from Colombia. IZA Discussion Paper 18963.
    Photo: David Wilson / Pexels