Your Interview Predicts 9% of Job Performance. Here's What Predicts the Other 91%
Hiring is the most consequential guess most organizations make. You commit a salary, a seat on the team, months of ramp time, and a bet on someone's future—and the centerpiece of that decision is usually a 45-minute conversation and a feeling.
Here's the uncomfortable part. When researchers measure how much a typical interview actually predicts about someone's future job performance, the answer is: not much. A conversation feels like insight. The data says it's mostly noise.
This isn't an argument that hiring managers are bad at their jobs. It's an argument that the main tool they've been handed—the unstructured interview—was never a very accurate instrument to begin with. It measures how someone performs in an interview, which is a completely different skill from how they perform in the job.
QUICK ANSWER
Research consistently finds that unstructured interviews explain only a small fraction of the variance in future job performance—by some analyses, under 10%. The interview mainly measures how well someone interviews, which is a different capability from doing the actual job. More predictive methods focus on observing a candidate perform job-relevant tasks—work samples, simulations, and job auditions—because they replace proxies (resume, degree, confidence) with direct evidence of ability. This shift from proxy to evidence is the basis of hiring simulations (R2R Select).
Key Insight : "Interviews explain roughly 9% of the variance in future job performance—meaning about 91% of what determines success is missed by the interview."— Psychology Today, on interview validity research
Hiring Runs on Proxies—and Proxies Only Cast a Shadow
Almost everything in a traditional hiring process is a proxy—an indirect signal standing in for the thing you actually care about:
The Resume is a proxy for capability.
The Degree is a proxy for knowledge.
The Previous Employer's Name is a proxy for quality.
The Interview is a proxy for performance.
The Confidence (speaking fluently about "driving impact") is a proxy for competence.
Proxies aren't useless—they narrow the field. But they indicate rather than evaluate, like trying to identify an object by looking only at its shadow. You can make a decent guess, but you can also be completely wrong because two very different objects can cast a similar shadow. A polished interviewer and a strong performer can look identical across a Zoom call.
"In one comprehensive analysis, out of 100 hiring decisions made on interviews, only about 8 accurately predicted on-the-job performance."— Schmidt & Hunter, as cited by ERE and Dr. John Sullivan
Why the Interview Specifically Rewards the Wrong Thing
The deeper problem is that interviews have become a rehearsed performance. Candidates can find the likely questions, prepare polished stories, and deliver answers engineered to sound impressive. The better someone is at preparing for and performing in interviews, the better they score—regardless of whether that skill has anything to do with the actual role.
So the interview quietly selects for interviewing ability.
For roles requiring smooth verbal performance under friendly pressure (certain sales and client-facing roles), that overlap might be useful.
For most other roles, it isn't. You end up optimizing for a skill you're not hiring for, and the cost of getting it wrong is steep.
"46% of new hires fail within 18 months—and the majority fail for reasons the interview was never designed to detect."— Leadership IQ
What Actually Predicts Performance: Evidence, Not Description
The methods that consistently outperform interviews share one trait: they stop asking candidates to describe their ability and start letting them demonstrate it. Instead of "tell me about a time you solved a hard problem," they hand the candidate a realistic version of a hard problem from the actual role and watch how they approach it.
This is the difference between a proxy and evidence:
Proxy: A story about handling an angry customer.
Evidence: Watching someone actually handle a simulated angry customer—how they open, when they concede, and how they recover.
This is exactly what hiring simulations do, and it's the basis of the R2R Select approach: start from the real role, identify what the job actually requires someone to do, and build a simulation that lets candidates produce observable evidence of whether they can do it—before the offer, not after. As covered in "Stop Asking Candidates to Describe the Job. Let Them Do It.", the shift from talking about the work to doing the work is what closes the 91%
gap.
Visual: Proxy vs. Evidence
Proxies (Indicate) | Evidence (Evaluates) |
Resume | Observed performance on a realistic task |
Degree & Past Employers | Real-time problem-solving under actual role constraints |
Polished Interview Answers | Direct demonstration of core skills |
Confidence & Fluency | Practical execution and decision-making |
Caption: "One casts a shadow. The other shows the object."
Frequently Asked Questions
Do job interviews accurately predict job performance?
Not very well. Research on interview validity finds that unstructured interviews explain only a small share of the variance in future performance—by some analyses under 10%. Interviews mainly measure interviewing ability, which is distinct from the ability to do the job, so they miss most of what determines success.
Why are interviews such weak predictors of performance?
Because they rely on proxies and rehearsed self-presentation rather than direct evidence. Candidates can prepare polished answers, and unconscious bias affects judgment, so the interview often rewards how well someone performs in a conversation rather than how well they'd perform in the role.
What predicts job performance better than an interview?
Methods that let candidates demonstrate ability on job-relevant tasks—work sample tests, job simulations, and structured assessments—consistently outperform unstructured interviews because they produce observable evidence of performance rather than a proxy for it.
What is the difference between a proxy and evidence in hiring?
A proxy is an indirect signal—a resume, a degree, an interview answer—that suggests capability without demonstrating it. Evidence is a direct observation of the candidate performing something close to the actual work. Proxies indicate; evidence evaluates.
How much does a bad hire actually cost?
Beyond direct replacement costs, a significant share of new hires—around 46% by one widely cited figure—fail within 18 months, carrying costs in lost productivity, management time, team disruption, and rehiring. Most of these failures stem from factors interviews don't reliably detect.
The R2R Select explainer breaks down why interviews reward interviewing—and how hiring simulations show what candidates can actually do.
Curious how this applies to your roles?
R2R Select is how we design hiring simulations that produce real evidence of candidate readiness—mapped to the specific roles you're hiring for. We walk organizations through how it works.
Comments