Assessment guidance
Skills, Not Résumés: Why the Document You’re Screening Was Never Built to Be Trusted
Part 1 of this series looked at why the interview itself is a weak predictor of job performance. This one goes back a step further, to the document that decides who gets an interview in the first place. A résumé is a candidate's own summary of their own career, written by someone whose entire incentive is to make it as compelling as possible. Treating that as a factual record was always a stretch. The data on how often it actually holds up is worse than most hiring teams assume.
Nirvana Intellect — Blog Series, Corporate Track, Part 2
Part 1 of this series looked at why the interview itself is a weak predictor of job performance. This one goes back a step further, to the document that decides who gets an interview in the first place. A résumé is a candidate’s own summary of their own career, written by someone whose entire incentive is to make it as compelling as possible. Treating that as a factual record was always a stretch. The data on how often it actually holds up is worse than most hiring teams assume.
The self-reported skill is doing more work than it can support
A background-screening benchmark from HireRight, cited by the Society for Human Resource Management, found that the large majority of employers who actually verify what’s on a résumé turn up a lie or a misrepresentation — a rate that has climbed steadily over the years as screening has gotten more common, not less (SHRM). The same reporting notes that even with that rate of discovery, only about half of employers bother verifying education credentials at all. Most hiring processes are, in effect, running on a document that’s wrong more often than not, checked by fewer than half the companies relying on it.
Dropping the degree requirement didn’t fix this
The obvious industry response over the last few years has been “skills-based hiring” — stripping degree requirements from job postings and evaluating candidates on skills instead. A joint study from the Harvard Business School Project on Managing the Future of Work and the Burning Glass Institute tracked what actually happened when large employers did this, and the results are sobering: for all the public commitments, the increased hiring opportunity for workers without degrees amounted to a small fraction of overall hires, concentrated almost entirely among a minority of employers who backed the policy change with real process changes (Harvard Kennedy School). Nearly half of the companies that announced the change made no measurable difference to who they actually hired. The degree requirement, in other words, didn’t disappear — it just moved somewhere less visible, often straight into an unstated preference the screening process still quietly enforces.
The pattern behind both findings is the same: removing a credential from a job posting, or trusting a claimed skill on a résumé, doesn’t create verified evidence of capability. It just removes one flawed proxy without replacing it with anything better.
What “verified” actually needs to mean
The Society for Industrial and Organizational Psychology (SIOP) — the professional body that sets the standard for employee-selection science — has been explicit that this evidentiary bar doesn’t lower just because a skill claim comes wrapped in a slicker format or an AI-generated match score. Its published recommendations for AI-based assessment set out requirements that any skills-verification method should meet: content clearly tied to the actual job, scores that are consistent and fair across candidates, evidence the score actually predicts job performance, and a documented trail of how the assessment was built and scored (SIOP). A self-reported bullet point on a résumé meets none of those bars. Neither, on its own, does an algorithmic “match score” that can’t explain its own reasoning.
This isn’t a technicality — it connects directly to the legal exposure Part 1 covered. The Uniform Guidelines on Employee Selection Procedures require that any selection method, including a skills screen, be demonstrably job-related and free of unjustified adverse impact (EEOC). A résumé keyword filter or an unvalidated skills test carries the same legal exposure as a bad interview — it’s just less visible because nobody thinks of a filter as a “decision.” It is one.
What a defensible skills-evidence process actually looks like
- Work-sample or scenario-based assessment tied to the actual tasks of the role, not a generic aptitude test standing in for job-specific skill.
- Scoring against a validated rubric, with evidence the score correlates with real job performance — not just face-valid plausibility.
- A documented chain from claim to evidence: what the candidate said, what they were actually tested on, and how the two connect — retrievable if a rejected candidate or a regulator asks.
- Independent human review of borderline or AI-flagged results, so a scoring model never becomes the final word on a candidate’s capability.
This is where AI is genuinely useful without becoming the liability: drafting structured evaluation of a work sample against a rubric, flagging skills claims worth independently testing rather than taking on faith, surfacing capability evidence from people already inside the organization as well as candidates outside it. A qualified person still signs off on what the evidence means for the decision. See how Nirvana Intellect governs skills evidence for hiring and workforce decisions →
The shift this actually requires
Neither of the two dominant fixes of the last decade — drop the degree, trust the résumé, or add an AI score on top of it — actually solved the underlying problem. The résumé was never built to be evidence, and removing one weak filter without replacing it with a validated one just relocates where the guessing happens. Explore Nirvana Intellect’s capability intelligence →
That’s Part 2 of this track. Next: why a single test score still can’t carry a hiring decision on its own, even once you’ve replaced the résumé with something real.
