Guide · 6 min read

How to decide what to build next

A method for choosing between competing ideas or features when everything looks plausible: comparison, opportunity cost, and one deciding question.

Having one idea is a question. Having five is a problem. When every option is defensible, the decision stops being about quality and starts being about comparison, and comparison only works if the options were judged the same way.

The real cost is not the build

Choosing what to build next is an opportunity-cost decision. The cost of the wrong choice is not the weeks spent building it. It is the weeks not spent on the option that would have worked, plus the momentum lost when the first one stalls.

This is why fast building raises the stakes rather than lowering them. When shipping takes a month, a bad choice costs a month. When shipping takes a weekend, you make twelve times as many choices, and the compounding cost of choosing badly goes up, not down.

Score them on the same axes or do not compare them

Two ideas evaluated in two separate conversations produce two incomparable answers. The framework has to be fixed before the options are. Use the same signals, in the same order, at the same depth, even when one option obviously excites you more.

Excitement is useful information about your own persistence. It is not information about the market.

Look at the shape of the score, not just the total

Two ideas can score identically and be nothing alike. One is evenly good everywhere. The other is exceptional in four places and broken in one. The second is usually the better bet, if the broken part is testable.

Evenly mediocre: usually a real business that is hard to distinguish.
Spiky with a fixable weakness: often the best option available.
Spiky with an unfixable weakness: a trap, because the strengths keep pulling you back.

The deciding question

When two options are genuinely close, ask which one you could kill fastest if you were wrong. Prefer the idea with the cheaper failure. Speed of disproof is a feature of the idea, not of your process.

Pick the option you can be wrong about most cheaply.

Then close the loop

Write the decision down with the reason attached, and revisit it when the next option appears. A portfolio of decisions with reasons becomes a personal pattern library, and you learn what you systematically overrate. A pile of abandoned projects teaches you nothing.