Skip to content
Download 1080 × 1080

Don't guess, measure what matters

Popular saying, in my words

Opinions are cheap and confident. A number you can check is neither, and it ends most arguments before they start.

Most technical debates are two guesses arguing. “The app feels slow.” “The new model is better.” “Users want this feature.” Each sentence sounds like a fact and contains none.

Measurement turns a debate into a decision. How slow, on which device, for which action? Better at what, on which test set, compared with what baseline? A number does not have to be perfect to be useful. It has to be defined, repeatable and connected to something a user would notice.

The second half of the line matters as much as the first. It is easy to measure what is convenient: lines of code, tickets closed, hours logged. Those numbers move without anything getting better. Pick the metric that would change your mind, then let it.

Try this

Take the next claim you hear in a meeting, including your own, and ask one question: what number would prove it wrong? If nobody can name one, it is still a guess.

engineering

Ask why until the root answers

The first answer to "why did it break?" is usually a symptom. Keep asking until the answer is a cause you can actually fix.

engineering

Ship for the outcome, not the output

Features, commits and releases are output. What changes for the user is the outcome, and it is the only part anyone pays for.

goals

A goal is a dream with a plan

A dream becomes a goal the moment it has a first step, a date and a way to tell whether you are on track.