Productivity rises sharply when system response drops below about 400 milliseconds, because the user stops waiting and starts flowing.
Doherty and Thadani's 1982 IBM study found the productivity gain from sub-400ms response paid for the hardware many times over. It is why perceived latency is treated as a feature rather than an optimisation.
In practice
- A search interface that returns results in 200ms is used exploratorily; the same interface at 1.5 seconds is used only when the user already knows what they want.
- A build that takes eight seconds gets run constantly; the same build at ninety seconds gets run when someone is confident.
- Subject
- Design & Systems
- Essay
- The limits every design runs into