Hypothesis
- NoteA small change in how users handle inbox organization can reduce repeat friction and improve follow-through.
Lab
Gmail Experiment
The lab focuses on what a good experiment needs: a real behavioral hypothesis, a defined segment, and a success metric that tells you whether the idea earned another iteration.
01
Experiment
Target segment
- NoteUsers who rely on Gmail heavily for work and need to manage high message volume with limited structure.
Test design
- NoteDefine the single behavior to influence before adding secondary metrics.
- NoteUse the simplest intervention that can still produce a measurable signal.
- NoteSeparate behavior change from opinion collection so the result is usable.
Success metric
- NotePrimary metric focused on behavioral completion, not surface engagement.
What I learned
- NoteA good experiment is mostly about choosing the right question.
- NoteSegments matter because the same intervention can mean different things across user groups.
- NoteMetrics should help you decide what to do next, not just look reasonable in a slide.