Inspect the Work, Not the Person
Judge people by their actual output, not by how they talk in a conversation.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 4
- Confidence
- —
Balfour argues it is nearly impossible to judge a person accurately from a conversation — interviews, performance reviews and promotion cases are all distorted by bias and by how well someone advocates for themselves. Instead, go to the work: review the artifacts a person has actually created or ship a simulation/exercise so you watch them do the job. At Reforge, 80–90% of the hiring decision is based on a portfolio review or a work simulation rather than the interview conversation. The same logic applies to promotions — surface the person's shipped work directly rather than relying on a manager's second-hand story, which is a game of telephone that loses talented but quieter people.
Origin
Balfour credits Frank Slootman for the phrasing (“inspect the work, not the person”); he says it named a principle Reforge was founded on but he lacked words for.
Core principles
- 01A conversation reveals how someone talks, not how they work.
- 02Portfolios and simulations show real capability and approach.
- 03Second-hand advocacy loses quiet, high-output people.
- 04Keep the work log lightweight or people won't maintain it.
How to run it
- 1
Replace the interview verdict with a work sample
Design the hiring loop so 80–90% of the decision rests on a portfolio review or a job simulation (an exercise that mirrors the real work), not the conversation.
Pro tip Reforge calls these 'simulations'; most companies call them exercises.
Watch out Your team will push back — building good simulations is real work and adds pipeline friction.
- 2
Interrogate the approach, not the answer
When reviewing the work, ask how they thought through it, what questions they asked, and what steps they took — you want conviction they have a process for similar-shaped problems, not the 'right' answer.
Pro tip You're testing transferable approach, not domain trivia.
- 3
Surface work directly in promotions
For promotion and performance decisions, let decision-makers look at the person's actual shipped work instead of relying on a manager's narrative, which invites bias and distortion.
Watch out Quiet high-performers with weaker self-advocacy get lost when you rely on the telephone game.
- 4
Keep a lightweight shipped-work log
Have people log what they helped ship and their role — a couple of screenshots and a few bullets, not an essay — emphasising work that reached customers.
Pro tip Log what shipped, not just 'wins' — good work can fail to win for reasons outside the person.
Watch out If logging is heavy, no one keeps it up and the signal disappears.
In the wild
Reforge bases 80–90% of every hire on a portfolio review or a work simulation. Balfour says every time he made an exception and skipped the simulation to reduce pipeline friction, he flat-out regretted it.
→ Higher-signal hires and fewer regretted decisions than conversation-led interviewing.
Lenny cites the extreme version — Linear runs a one-to-five-day work trial where a candidate actually works with the team — as the same principle taken to its limit.
→ The team observes real collaboration and output before deciding.
Common mistakes
Trusting the interview conversation
Drawing a hire/no-hire conclusion from a conversation loads the decision with bias and rewards smooth talkers over strong operators.
Making the work log heavy
If capturing work takes an essay every two weeks, people stop doing it and you lose the very signal the system depends on.
Is it for you?
Best for
Founders and managers designing hiring loops or promotion processes.
Not ideal for
Roles where no reviewable artifact or work sample can exist.
From the episode
Brian Balfour: 10 lessons on career, growth, and life