Has Linear Ever Run Split Tests?

Karri Saarinen says Linear has never run split tests. Instrumentation is used only when a specific question needs an answer. Watching the dials and looking out the window is probably better, while many companies pick one and more often trust the dials. Soleio compared that habit to a driver following Google Maps off a cliff, and Saarinen agreed.

Linear is a collection of many workflows. Weekly active users and other counts are tracked, but a new feature does not have to be used by everyone, or even by every company in a segment. Company needs are similar and also different by size and culture, so one success metric is a poor fit. Feedback is collected in Linear from discussions, sometimes automatically and sometimes by hand. When a larger company enters a pilot, the team asks about processes and problems, then keeps talking after they start using Linear. Problems nobody knew were there get fixed or improved, on the view that a fix for one customer will scale to other customers.

AI features are where usage should be monitored more. The point of watching is that the numbers will tell the story of what is actually valuable or meaningful.