# A/B Test Launch Checklist

Use this checklist before an experiment goes live.

## Experiment Setup

- The business problem is clearly defined.
- The hypothesis is measurable.
- Control and treatment versions are documented.
- Target users are defined.
- Randomization method is documented.
- Traffic split is agreed.

## Metrics

- One primary metric is selected.
- Guardrail metrics are selected.
- Metric formulas are documented.
- Minimum detectable effect is estimated.
- Sample size is calculated.
- Experiment duration is planned.

## Data Quality

- Event tracking has been tested.
- Exclusion rules are documented.
- Bot, test, and internal traffic rules are defined.
- The analysis owner is assigned.
- The decision rule is written before launch.

## Decision Rules

- Launch if the primary metric improves and guardrails are stable.
- Continue observing if sample size is insufficient.
- Roll back if guardrails degrade materially.
- Re-run or redesign if tracking or randomization fails.