Hiring, signal vs noise

Hiring is the single highest-leverage activity in a company, and the one most managers do worst. The cost of a bad hire is 3-5x their salary when you count ramp, lost output, team drag, and eventual severance. Yet most interview processes measure charisma, not performance.

What predicts job performance (validated research)

Meta-analyses of hiring-success predictors, ranked by correlation with actual performance:

• Work sample tests, having them do the actual job (r ≈ 0.54)
• Structured interviews, same questions, scored rubric (r ≈ 0.51)
• Cognitive ability tests, r ≈ 0.51
• Integrity tests, r ≈ 0.41
• Reference checks (done well), r ≈ 0.26
• Years of experience, r ≈ 0.18 (weak)
• Unstructured interviews, r ≈ 0.14 (shockingly weak)
• Age, r ≈ 0.01 (no predictive value)

What most companies do: unstructured interviews + vibe. What they should do: work samples + structured interviews.

The core principle: predict work by watching work

The best interviews are ones where the candidate does something that looks like the actual job. An engineer pairs on real code. A marketer drafts a real email. A salesperson runs a simulated call. The signal is enormous compared to "tell me about a time when you…"

The hiring process that works

Stage 1: Role scorecard (before posting)

What does success look like in this seat after 90 days? 1 year? If you can't write this in 1 page, you're not ready to hire. See The role scorecard.

Stage 2: Sourcing

Inbound only = slow + biased toward people who see your post. Best hires are often passive candidates sourced directly. Invest in sourcing, not just application review.

Stage 3: Screen (20-30 min)

Recruiter or hiring manager. Goals: confirm baseline fit, compensation alignment, motivation. Avoid elaborate questions, this is a filter, not an interview.

Stage 4: Hiring manager interview (45-60 min)

Structured. Same questions for every candidate. Scored rubric. Topics:

• Deep-dive on a past project relevant to the role
• Situational: "here's a scenario, walk me through how you'd approach it"
• Tradeoff questions that reveal thinking style

Stage 5: Work sample (60-120 min)

Paid, always. Something concrete they'd actually do in the job. Bonus: do it together so you see their working style, not just the output.

Stage 6: Panel interviews (60 min each)

3-5 people from the team. Each person owns 1-2 specific dimensions. Post-interview, each writes their evaluation BEFORE talking to anyone else. Group discussion after, no anchoring.

Stage 7: References

Do them. Back-channel even more than front-channel. Ask specific questions: "On a scale of 1-10, where was [candidate] in your team?" "Would you hire them again?" "What was the hardest thing for them?"

Stage 8: Offer

Fast. Move decisively. A week to decide + extend beats 3 weeks of deliberation. Top candidates have options; speed matters.

The "no hire" is cheaper

When the team is split, default to "no hire." A bad hire costs 3-5x salary. A missed good hire costs.., you keep looking. The asymmetry is massive.

Interview questions that actually reveal something

• "Walk me through the most complex project you've run, end to end."
• "Tell me about a conflict with a peer. What did you do?"
• "What's a decision you made that you later realized was wrong?"
• "Who are the 3 best people you've worked with? What made them great?"
• "If I called [specific former manager], what would they say are your strengths and weaknesses?"

Red flags

• Blames past employers / teammates for problems
• Can't describe their impact in concrete numbers
• Over-polished answers (rehearsed)
• No mistakes in their work history (evasion)
• Over-indexed on credentials vs, outcomes
• Won't share specifics (confidentiality excuse is sometimes real, often a dodge)

Green flags

• Describes failures + what they learned
• Credits team members for wins
• Asks substantive questions about the role + business
• Articulates tradeoffs + uncertainty
• Can describe their impact in numbers
• References check out with specifics

The most underrated signal

How do they treat the receptionist, coordinator, and anyone not in the interview? The real personality shows up when nobody's "watching."

What to do with this

• Invest in work-sample tests first, interviews without work samples produce high noise + bias
• Get backchannel references, not just listed references, candor comes from mutual connections not official contacts
• Ask past-behavior questions, "tell me about a time" outperforms hypothetical "what would you do"
• Skip brainteasers, they correlate poorly with job performance and select for a narrow skill
• Watch how candidates treat everyone, personality reveals when they don't think they're being evaluated