02 — Capability · BC-1260.30
Experimentation
Run controlled experiments on real users, read the results with honest statistics and act on them, so the product learns from what people do rather than what they say.
- A/B Testing
- Controlled Experiments
- Experiment Platform
In scope
- Experiment design and hypothesis review
- Randomization and exposure tracking
- Statistical analysis and decision rules
- Guardrail metrics that stop a harmful test
Out of scope
- The flag mechanism used to split traffic (see BC-1220)
- Qualitative user research (see BC-410)
Realized by · 0
- No product in the catalog yet.
Used in · 1
Build it · 1
- pattern Feature Flags via BC-1260.30.20
Decomposes into · 4
- BC-1260.30.10Experiment DesignState the hypothesis, the primary metric, the minimum effect worth detecting and the sample it needs before the test starts.
- BC-1260.30.20Assignment & Exposure TrackingRandomize users or tenants into variants and record exactly who saw what, so the analysis is on real exposure.
- BC-1260.30.30Statistical Analysis & ReadoutAnalyze results with pre-registered methods, report uncertainty plainly and resist calling a winner early.
- BC-1260.30.40Guardrails & Experiment ReviewDefine metrics that must not degrade, stop tests that breach them and review experiments for ethics and overlap.