The "I don’t do interviews" approach replaces traditional interviews with real work — paid trial projects, actual tasks, or work samples — to evaluate candidates. It exists because interviews primarily measure interviewing ability (composure, polish, rehearsed answers), which correlates weakly with job performance, while watching someone do representative work measures the thing you actually care about.
- ▪Interviews mostly measure how well someone interviews.
- ▪Interview skill correlates weakly with job performance.
- ▪Trial work and real tasks measure actual ability.
- ▪You watch the work instead of hearing about it.
- ▪Hire for what the job needs, not for polish.
The traditional interview is a strange ritual when you examine it: you put a candidate in an artificial, high-pressure conversation and judge their fitness for a job by how smoothly they answer questions about themselves. What that reliably measures is interviewing skill — composure, rehearsed stories, charisma under pressure — which is a real talent and almost never the talent the job requires. Great interviewers can be mediocre performers, and excellent workers can interview badly. The signal and the thing you care about barely overlap.
The fix is to stop asking people to describe their work and start watching them do it. A paid trial, a representative task, a real work sample — these measure the actual job, not the performance of being interviewed.
Interview vs. real work
The two evaluate fundamentally different things, only one of which is the job.
| Interview | Trial work | |
|---|---|---|
| Measures | Interviewing skill | Job ability |
| Evidence | Self-description | Actual output |
| Favors | Polish, charisma | Competence |
| Predicts performance | Weakly | Strongly |
Why interviews mislead
Interviews are vulnerable to exactly the wrong things: the confident candidate who interviews beautifully and underperforms, the brilliant one who freezes in conversation, the rehearsed answers that reveal preparation rather than ability, and the interviewer’s own biases toward people who present like them. You end up selecting for a skill — performing in interviews — that you’ll rarely need again after the hire. The method optimizes for the audition, not the role.
Relative predictive value of each method.
How to evaluate with real work
The practical version: give candidates a paid, representative task — real or realistic work they’d actually do — and evaluate the output and the process. Pay for their time (it’s respectful and gets serious effort), keep it scoped and fair, and judge it against what the role genuinely requires. You learn more from one real task than from an hour of polished conversation, because you’re finally measuring the job instead of the candidate’s ability to talk about it.
Don’t you still need to talk to candidates?
Hiring is too consequential to decide on a performance that measures the wrong skill. Replace interview theater with real work — paid, representative, judged against the actual role — and you select for the thing you’re actually buying: the ability to do the job, not the ability to talk about doing it.