Do Half
A principle from Scott Belsky — a deliberate discipline of building fewer features, serving a narrower market, and eliminating existing functionality in order to concentrate usage on the core action that drives the business.
The formulation: only build half the features you plan to. Offer half the options. Target half the market. Not as a resource constraint, but as a product strategy.
The Behance discovery
Belsky discovered this principle empirically at Behance. The team had built several secondary features — a Tip Exchange, Groups, portfolio colour controls. Killing the Tip Exchange caused project publishing (the core metric) to increase. Killing Groups caused it to increase again. ‘If you make the whole product about one thing, everyone does that one thing at 10× the velocity.’
The pattern generalised: each removal concentrated user attention on the action that drove traffic and business value. The inverse — addition — typically diffused it.
The operating principle: optimise for the problems you want
Belsky distinguishes between two categories:
- Brick walls: catastrophic failure modes that must be eliminated (security failures, crashes, experiences that cause users to leave permanently).
- Problems you want to have: scale problems, retention problems, growth problems. These are signs the product is working.
The Do Half discipline applies to the second category: when choosing which features to build or kill, select based on which choice brings you closer to the problems you want to have.
The 24-hour test
When Behance removed portfolio colour controls, users complained for 24 hours. After that, nobody mentioned it again. The portfolios looked cleaner; users published more projects. This asymmetry — loud short-term objection, invisible long-term gain — is characteristic of well-chosen cuts.
Where mainstream views differ
The conventional product instinct is to add. Users request features; roadmaps grow; ‘just one more thing’ is the default response to every customer conversation. Do Half inverts this: the question is not ‘what should we add?’ but ‘what can we safely remove to drive more of the one thing that matters?’ This is counterintuitive because every individual removal generates visible objections while the benefit — increased core-action frequency — is diffuse and delayed.