Experiment design
Choosing Guardrail Metrics for a Feature Experiment
A useful guardrail protects customer or business health without turning every experiment into an impossible approval process.
Read field note ↗Methods for product teams
App experiment field notes
Concise guidance on designing, checking and interpreting feature experiments.
Data quality
Why Experiment Exposure Is Not the Same as Feature Assignment
Analysing everyone assigned to a variant can conceal the effect among customers who actually encountered the feature.
Read field note ↗Decision practice
A Practical Experiment Review for Product Leaders
A review meeting should produce a decision and a learning record—not a tour through every chart.
Read field note ↗