Experiment velocity

Experiment velocity is the rate of running tests. High velocity compounds learning over time. Low velocity means slow progress.

Typical velocity

• Early stage: 1-2 tests per week
• Mature growth team: 5-10+ per week
• Top teams: 20+ per week

Building velocity

• Templates for common test types
• Experimentation infrastructure (Optimizely, Split, LaunchDarkly)
• Backlog of hypotheses
• Regular planning cadence

Quality vs quantity

Bad tests at high velocity is wasted effort. Good hypotheses + discipline beat noise.

What to do with this

• Aim for 3-5 tests per week, not one heroic test per quarter, compound of small wins beats rare big ones
• Treat every test as a learning, not just a pass/fail, even inconclusive tests narrow the next hypothesis
• Reduce cycle time, automation (feature flags, test tooling) removes friction between "idea" and "live test"
• Quality + velocity are both required, 200 random tests produce noise, 20 disciplined tests per quarter beat both
• Track velocity as a leading indicator, teams shipping 1 test per week usually outperform teams shipping 1 per month by 3-5x annually