experimentation

To design and run A/B tests with pre-registration and guardrails, you create a hypothesis document, perform power calculations, and define explicit ramp planning and abort rules before launching the experiment.

An SRM check in experimentation detects sample ratio mismatch by comparing observed allocation against expected proportions. You need it during analysis to validate that feature flag ramping did not skew your experiment data.

Calculate sample size and power for feature flag experiments by defining your hypothesis and expected effect size upfront. This Skill enforces pre-registration of these calculations to ensure measurable, valid product decisions.

The best way to manage feature flag lifecycle and cleanup is using a structured post-experiment process. This Skill provides documentation templates and cleanup workflows to address tech debt and record learnings.

Yes, you can use this workflow for hypothesis-driven product decisions without external dependencies. It operates independently using markdown documentation and YAML manifests to enforce reproducibility and clear ownership.

Experiment results often lack reproducibility across teams due to missing pre-registration and documentation. This Skill enforces a repeatable workflow with explicit decision rules and flag manifests to ensure consistent learnings.