Design an A/B test measurement plan with Claude Code

Use a-b-test-design-pm to turn a product hypothesis into metric definitions, a baseline, a target, and a review plan.

Install in Claude Code

Claude Code, a product hypothesis, and baseline counts.

Open the pinned GitHub source below and retrieve the `product-manager/analytics/skills/a-b-test-design-pm/` directory from that revision.
Copy it into `.claude/skills/a-b-test-design-pm/` in your project, keeping `SKILL.md` and its companion files together.
Open Claude Code in the project root and enter `/a-b-test-design-pm` followed by the example task.

Example task

Use 200 form starts and 80 submissions over 14 days as sample data. Plan a test of a three-step form with a 45% completion target. Ask for the denominator, tracking events, and review cadence.

What to check

The plan records the 40% baseline, 45% target, and one completion formula for both groups.

Original SKILL.md on GitHub · Updated 2026-11-11

Frequently asked question

Can the old completion rate establish the impact of the new form?

It supplies the baseline. Evaluate impact using comparable group data and the assumptions of the experiment.