What this Skill helps with

Ad and Creative Test Designer turns a campaign question into a test plan before launch. It defines the hypothesis, control, variant, primary metric, guardrails, sample requirement, duration, and stopping rule so the result is easier to interpret.

Who this Skill is for

It is intended for performance marketers, creative strategists, growth teams, and product marketers planning ad, landing-page, email, or interface tests.

What you provide

  • The decision the test should inform
  • Baseline traffic, conversion rate, cost, or event volume
  • The minimum useful change and available budget
  • Proposed control and variants
  • Platform, audience, placement, and campaign constraints
  • Dates, promotions, or other factors that may affect the result

What it produces

  • A clear test hypothesis and decision rule
  • Control and variant definitions with one isolated variable where possible
  • Primary metric, guardrails, and diagnostic metrics
  • Sample-size and duration assumptions
  • Platform setup notes and a pre-launch checklist

How the Skill works

The workflow starts from the business decision, then checks whether the proposed test can answer it. It flags mixed variables, weak baselines, overlapping audiences, short durations, and stopping rules that could bias the result.

Installation and first use

  1. Download and add the Skill to Claude.
  2. Describe the decision, baseline, and available traffic or budget.
  3. List the control and proposed variants.
  4. Ask for assumptions and feasibility concerns before the final plan.
  5. Confirm platform setup and measurement with the person running the test.

Example workflow and expected output

Provide the baseline conversion rate and weekly traffic for a Meta landing-page test comparing two offers. Ask for a hypothesis, sample assumption, duration range, audience split, guardrails, and a result table to complete after the test.

Limits and review guidance

Sample estimates depend on accurate baselines and statistical assumptions. Platform delivery may not split traffic evenly, and outside events may affect results. A qualified analyst should review high-cost or high-impact tests.

Frequently asked questions

Can it design multivariate tests?

Yes, but it should first check whether traffic and budget are sufficient for the number of combinations.

Does it choose the minimum useful change for me?

It can suggest a range, but the business owner should define what difference would justify action.

Does it analyze completed results?

This Skill plans tests. Use a results-analysis workflow after the test finishes.