DhungJoo Kim.
← Back to memo
DK · Memo · Selective take
001· issue
Filed · 04·17·26Read · 4 minSelective take

Hiring isn't the fix. The spec is.

Nine of the last twelve hires across the portfolio that didn't stick failed at the spec, not the person. Here's the thing I write before the job post now.


Situation

I spent a quarter convinced we had a hiring problem. We'd brought in nine people across the portfolio in fourteen months, three of them left, three of them we let go, and three of them became permanent. A 33% stick rate is fine for a single role. Across a portfolio, it's a process problem.

I went back through the six that didn't stick. I was looking for a pattern in the people. What I found was a pattern in the briefs.

Decision

Before any new hire — full-time, fractional, contractor — I now write a one-page document. Five lines, in this order:

  1. The artifact. What the role produces, weekly. Not "growth" — the weekly report, the email sequence, the campaign brief.
  2. The metric. One number that goes up if this role is working. Not three. One.
  3. The check-in. When and how I'll know it's working before the 90-day mark. Usually a 30-day artifact review.
  4. The failure mode. What this role looks like when it isn't working. Most of mine looked like busy person, no artifact.
  5. The replacement. If this person leaves in six months, what's the next move. If I can't answer this, the role isn't real yet.

If I can't write all five in a sitting, I don't post the role. The spec is the work. Recruiting against an unfinished spec just gets me a better-shaped version of the wrong shape.

Insight

We don't have a delegation problem. We have a specification problem. The hire is a downstream of the spec. The spec is the moat.

— DK


DK
Filed · 04·17·26
Memo · Issue 001 of 3
Operator log · Cool Forest · v1.0
Subscribe

Get the next memo on Monday.

One a week. Filed at 8am ET.

Next →
002· issue
An image-prompt pipeline in an afternoon.