The interview is the ritual we trust most. It’s also one of the weakest predictors of who will actually do the job. A practitioner’s case for scrapping the game.

Picture a young engineer sitting across the table from you. He sleeps in the office. He talks, with a straight face, about dying on Mars. He is blunt to the edge of rude and has no patience for small talk. Every instinct you have in that room says no. Now imagine his name was Elon Musk, years before anyone knew it.

You would not have hired him. Be honest. On the strength of a forty-five-minute conversation and a gut read, almost none of us would have.

I’ve sat on both sides of that table for close to two decades — as the person being sized up, and as the person doing the sizing — and I’ve landed on an uncomfortable conclusion about the whole exercise. Interviewing is a selfish act. We tell ourselves we’re measuring competence, judgment, whether someone can carry a hard role under pressure. What we’re actually answering, underneath the templated questions, is one private thing: How would it feel to sit next to this person every day?

That question is not illegitimate. But it is not the same as is this person good at the work — and we routinely let the first one quietly answer the second. Even a genuine talent gets graded on how pleasant they seem, how smoothly they sell, whether some small thing about them clicks with you personally. Drop a real-life Dr. House into your hiring funnel and watch the bedside manner sink him before anyone gets near the diagnosis he’d have nailed.

A verdict reached in minutes, dressed up as an hour

The most honest thing about a job interview is how fast the mind makes up its mind. Research on first impressions shows we form judgments about a stranger’s trustworthiness and competence within seconds of seeing their face, before they’ve said a word. In the interview room, studies suggest a large share of interviewers reach a verdict inside the first five minutes. What happens next isn’t evaluation — it’s confirmation. Once the gut has ruled, the rest of the hour becomes a hunt for evidence that we were right. Psychologists call it confirmation bias. Hiring managers call it a good conversation.

So the question worth asking is simple: does any of this actually predict who will be good at the job? For decades the reassuring answer came from a landmark 1998 meta-analysis by Frank Schmidt and John Hunter, which pooled 85 years of research and crowned the structured interview and cognitive ability as the strongest tools we had. Then in 2022, a team led by Paul Sackett at the University of Minnesota went back and corrected a statistical error that had quietly inflated those numbers for a generation. The revised picture is sobering. The structured interview — the disciplined kind with a fixed script and a scoring rubric — came out as the single best predictor, and even it explains only about a fifth of the variation in how people actually perform. The unstructured interview, the freewheeling “let’s just talk and see if there’s a fit” conversation that the overwhelming majority of companies actually run, sits near the bottom of the list.

Sit with that. The centerpiece of nearly every hiring decision on earth is a method the best available science ranks as barely better than noise.

Three tools, all of them cracked

If not the interview, then what? The instinct is to fall back on the other two pillars of hiring. Both buckle the moment you lean on them.

The resume gets, on average, about 7.4 seconds of human attention — that’s the figure from an eye-tracking study by the career site Ladders, watching real recruiters skim real resumes. Seven seconds to decide a stranger’s professional worth. And what they’re skimming isn’t always true: depending on which survey you trust, somewhere between a quarter and nearly two-thirds of workers admit to stretching the truth on a resume, and one 2024 study of people who used outright fraudulent resumes found that 63% still landed an offer and 96% were never caught. Meanwhile the applicant-tracking software meant to impose order routinely filters out perfectly qualified people for using the wrong keywords. The document is either padded, misread, or both.

The reference check is the fox guarding the henhouse. The candidate hands you the list. Nobody in the history of employment has ever submitted a reference primed to sink them. You are, by design, calling people selected specifically to say nice things.

The interview we’ve already covered. It’s the weakest of the three, and it’s the one we weight most heavily.

Calling references doesn’t work, because the candidate chooses them. Reviewing resumes doesn’t work, because they can lie and the good ones drown anyway. Interviewing doesn’t work, because it’s too short to see the truth and too easy to fall in love with the wrong person. Strip those three away and look at what’s left holding up the entire apparatus. There’s nothing there.

The receipts

None of this would matter if the failures were rare. They aren’t, and the bill is enormous.

The U.S. Department of Labor puts the cost of a single bad hire at a conservative 30% of that person’s first-year earnings — and that’s just the floor, covering the obvious replacement costs. The Society for Human Resource Management pegs the real number anywhere from half to twice an annual salary, climbing toward 200% for senior roles. CareerBuilder’s survey work lands the average loss around $17,000 per bad hire, and up to $240,000 for a specialized or executive miss. Roughly three in four employers admit they’ve hired the wrong person. Gallup, tracking the downstream drag of disengaged employees, estimates the annual cost to U.S. companies at $450 to $550 billion.

Then there’s the number that should end the debate. Leadership IQ spent three years studying more than 20,000 new hires in its Hiring for Attitude research. It found that 46% of them failed inside eighteen months, and only 19% went on to unambiguous success. Fewer than one in five. And the failures weren’t about talent — 89% traced back to attitude and interpersonal fit, only 11% to a shortfall in technical skill.

Here’s the part that indicts the process rather than the people: 82% of the hiring managers said that, looking back, they’d seen the warning signs during the interview. They saw it coming. They hired anyway. The signal wasn’t missing. The process simply had no way to act on it — because a warm room, a firm handshake, and a candidate who sells well can drown out almost anything.

I can’t reliably tell how good someone is at the work until I’ve worked beside them. Neither can you. Anyone who claims they can read it off a one-hour interview with anything like certainty is describing a confidence they have, not an accuracy they possess. Every practitioner I know has watched a “sure thing” collapse and a quiet, unremarkable candidate turn into the best hire on the team. The surprise runs in both directions, every time, which is the whole problem.

The part where we fix it

So what would actually work? Not a better list of questions. Not another AI-generated question bank layered on top of a broken ritual. The rot is structural, and structural problems need structural answers.

Start with the thing that already exists but stays locked away: honest, firsthand knowledge of how someone actually works. Right now the only people who truly know a candidate’s value are the ones who’ve already worked with them — and that knowledge dies inside old Slack threads and the private memories of former managers. Imagine a network of companies that contributes verified reviews of the people it has actually employed, with the same obligation to leave an honest record that a candidate has to be honest on an application. Not curated references. Not a LinkedIn endorsement traded between friends. A real, accountable track record, built by the people who watched the work happen.

Pair that with a shift on the candidate’s side. Stop the exhausting theater of blasting a hundred tailored resumes at a hundred overwhelmed recruiters. Let a person maintain one living, portable record of their real experience and skills — a lifelong profile in something as simple and durable as a plain and extensive text file with any work related data — that a company’s agents can query directly for exactly the capabilities and the verified history the role needs. The candidate stops performing. The company stops guessing. The match gets made against evidence instead of vibes.

The pieces are already on the table. Verified credentials exist. The whole industry is inching, fitfully, toward skills-based hiring and away from proxies like degrees and titles. Portable professional records are technically trivial. What’s missing isn’t the technology. It’s the will to stop playing a dating game everyone privately admits is broken.

If you work in hiring and you’ve seen what I’ve seen — the brilliant people passed over, the confident mediocrities waved through, the whole thing riding on a gut feeling formed in the first five minutes — then you already know the ritual isn’t sacred. It’s just old. The tools finally caught up to the problem. Someone needs to build the thing that replaces it.

It’s time for a revolution in this field. The ideas are sitting right here, waiting their turn.

---

Sources

• Cost of a bad hire: U.S. Department of Labor (30% of first-year earnings); Society for Human Resource Management, Human Capital Benchmarking research (50–200% of salary); CareerBuilder survey (~$17,000 average; up to $240,000 for specialized/executive roles; ~74% of employers report a bad hire).
• Cost of disengagement: Gallup (estimated $450–550 billion annually in lost U.S. productivity).
• Predictive validity of selection methods: Schmidt, F. L., & Hunter, J. E. (1998), Psychological Bulletin; Sackett, P., Zhang, C., Berry, C., & Lievens, F. (2022), “Revisiting Meta-Analytic Estimates of Validity in Personnel Selection,” Journal of Applied Psychology (structured interviews as the strongest single predictor at roughly r = .42; unstructured interviews near the bottom of the hierarchy).
• First impressions and interview timing: research on rapid face-based trait judgments (Willis & Todorov and related work); Frieder, Van Iddekinge & Raymark (2015) on how quickly interviewers reach decisions.
• Resume screening and honesty: Ladders eye-tracking study (~7.4 seconds per resume); StandOut CV survey (64.2% admitted stretching the truth on skills, experience, or references); Resume.org 2024 survey (of those using fraudulent resumes, 63% received offers and 96% were never caught). Reported rates vary widely by survey and question wording.
• New-hire failure: Leadership IQ, Hiring for Attitude study (46% of new hires failed within 18 months; 19% achieved unequivocal success; 89% of failures attributed to attitude/interpersonal factors vs. 11% technical; 82% of hiring managers reported seeing warning signs during the interview).