← All issues

The Compound Years · Issue #16 · October 2026

The half-hour interview that costs years

The first thirty minutes of an interview do not screen for what most people think they screen for.

01The experiment

Google ran the numbers on its own interviews and found four rounds told them almost everything a dozen did.

Google’s people analytics team looked back at its hiring data and found interviewer scores converged after about four conversations. The fifth onwards added very little, and the famous brainteasers predicted nothing at all. They published it and cut their own process. A half-hour interview is a long time to spend on a signal that turns out not to be one.

02The lesson

You only learn to read an interview by sitting in on enough of them to build the pattern.

Reading hiring books gives you vocabulary. Sitting through fifty interviews gives you the pattern recognition. The dataset I trust most about hiring is the one I built by being in the room -- watching which ”great on paper” candidates fell apart on the first specific follow-up.

03The founder

Reed Hastings built the Netflix ”keeper test” into every interview at the company.

Hastings’s discipline was simple: would you fight to keep this person if they tried to leave? If not, they should not be in the role. Netflix bakes that question into the interview itself -- and the half-hour conversation is structured to surface whether the answer is genuinely yes.

04The mindset

Hiring is the most reversible-looking, irreversible-feeling decision you make.

You can technically let someone go. The cost -- to the team, to the work, to your own time -- is real. The habit I am building is to slow down for that one decision in a way I would not for almost any other call on a Tuesday.

05The question

What is the one signal you would pay closest attention to in the first ten minutes of an interview?

Hit reply and tell me. I read every one.


Forward to one person about to make a hire in the next thirty days.

-- David