My Writings

Lessons from two years of building product

Mostly things I got wrong first.

Always question what the numerator and the denominator actually mean

Owning a KPI starts with refusing the definition you were handed, and cutting through the noise around it.

Users perceive value through design, so good is not good enough

Hierarchy and clutter are the easy part. Alignment, adjacency and flicker are what people actually feel.

What to keep upfront, and what to put behind a click

Let people skip the boring parts quickly. This is not academia — nobody likes reading more than three words.

Start with the customer problem, then work back to the tech

A big problem may justify a new vendor contract. A small one doesn't deserve an hour, let alone an AI copilot.

In a crisis, leadership doesn't want a gap analysis

They want radical action, a clear problem statement and an impact number.