Skills-based hiring is the practice of selecting candidates on demonstrated ability — measured through assessments, work samples, and verified credentials — rather than proxies like degrees, pedigree, or years of experience. By that definition, most organisations are not doing it. They have adopted the language, rewritten the job ads, and left the actual selection mechanics untouched. The result is a widening gap between what employers say about skills-based hiring and what their hiring data shows — and 2026 is the year that gap became impossible to ignore.
The 85% vs 0.14% problem
Surveys keep telling one story. In TestGorilla's research, around 85% of employers say they use some form of skills-based hiring. Job postings tell another: analysis by Harvard Business School and the Burning Glass Institute found that at companies which publicly removed degree requirements, the share of new hires who actually lacked a bachelor's degree rose by just 0.14 percentage points. For every 700 roles "opened up" to non-degree candidates, roughly one such candidate was hired.
The trend line has since turned the wrong way. Indeed Hiring Lab data shows the share of US postings requiring a bachelor's degree or higher climbing again — from 16.6% in late 2023 to 19.3% by November 2025 — and, more tellingly, rising within the same job titles, not just because of a shift toward white-collar roles. Employers who loudly dropped the degree filter are quietly putting it back.
It is tempting to read this as hypocrisy. The more useful reading is that it is an infrastructure failure. Removing a filter is a policy decision; replacing it is an engineering problem, and most organisations only did the first half.
What the degree was actually doing
A degree requirement is a crude instrument, but it performed three jobs at once. It signalled that the candidate completed something long and difficult. It implied a baseline of domain exposure. And it outsourced vetting to an institution the employer did not have to pay or manage.
Strip the degree out and a recruiter is left staring at self-reported claims: skills lists on a CV, endorsements on a profile, "proficient in advanced Excel" with nothing behind it. Faced with 300 applications and no way to verify any of them, reverting to the degree filter is not prejudice — it is the only remaining lever that costs nothing to pull. That is how you get 85% stated intent and 0.14% behavioural change: the old proxy was deleted before a new signal existed.
AI just burned down the last of the old signal
Whatever information a polished CV still carried, the generative-AI application wave has destroyed it. Applications per hire have roughly tripled since 2021 to more than 300 per open role; LinkedIn was reporting around 11,000 applications submitted per minute, up more than 45% in a year. Nearly six in ten hiring managers say they now routinely encounter AI-generated resumes.
This is usually framed as candidate misbehaviour. It is better understood as a rational response to automated screening: employers built keyword-matching machines, so candidates built keyword-generating machines. The arms race has a clear loser — the resume itself. When every application is fluent, tailored, and confident, fluency stops discriminating between candidates at all.
That collapse has an underappreciated consequence for skills-based hiring: it removes the option of doing nothing. An employer could coast on CV-plus-interview when applications carried some honest signal. At 300 indistinguishable applications per role, the only screening signals left standing are the ones a candidate cannot generate with a prompt — proctored assessments, structured work samples, and credentials issued by someone other than the candidate.
What replacing the filter actually takes
Organisations that have closed the say–do gap tend to have built three things.
1. A demonstrated-skill gate early in the funnel
If the first real filter in your funnel reads documents, you are still hiring on claims. The fix is to move a short, job-relevant assessment to the top of the funnel — before human review, applied to everyone above a minimal bar. Structured scenario tasks and situational judgement items measure the thing the job actually requires, and modern AI grading makes it economic to score open-ended responses at screening volume rather than reserving them for finalists. Integrity matters at this stage precisely because the assessment is unsupervised: platforms like AssessAll pair AI-graded scenario responses with AI proctoring that reports an integrity band alongside the score, so a screening result arrives with its own trust label attached.
2. Verifiable credentials instead of self-assertion
The second component is portability: a skill demonstrated once should not evaporate after one hiring process. Standards such as Open Badges 3.0 and the W3C Verifiable Credentials model now make it possible to issue a signed, machine-checkable claim — this person, this skill, this evidence, this issuer — that a third party can confirm in one click. This is the direction Skill Passports point in: an assessment result that belongs to the candidate and can be presented to the next employer as evidence rather than assertion. For recruiters, a verified credential restores exactly what the degree used to provide — third-party vetting — without the four-year proxy.
3. Economics that allow testing everyone
The quiet reason skills-based hiring stalls is unit cost. Enterprise assessment contracts priced per seat or per annum push organisations to test late and test few — which reinstates the CV as the de facto first filter. Screening on demonstrated skill only works if assessing 400 applicants is an unremarkable expense. This is where pay-as-you-go models change behaviour: at AssessAll's credit pricing of about ₹30 (US$0.50) per assessment, testing an entire applicant pool costs less than one recruiter-day of CV screening — and new accounts get free credits (100 for individuals, 250 for corporate) to trial the funnel before committing anything.
The honest scorecard
If you want to know whether your organisation practises skills-based hiring or merely endorses it, four questions settle it. What share of candidates complete a scored, job-relevant task before a human shortlists them? Can a hiring manager see verified evidence — not claims — for the top three skills in the role? Has the share of hires without the previously "required" background actually moved? And when hiring volume spikes, does assessment coverage hold, or does the team fall back to eyeballing CVs?
Most organisations fail at least three. That is not a reason for cynicism about the skills-first idea — the Harvard/Burning Glass work also found that when non-degree hires are made through genuine skills evaluation, they perform comparably and stay longer. The idea works. The implementation was never built.
Takeaway
Skills-based hiring did not fail; it was announced without being installed. The employers who close the 85%-say/0.14%-do gap in 2026 will be the ones who replace deleted proxies with real infrastructure — assessment at the top of the funnel, integrity signals attached to every score, and credentials a stranger can verify in a click.