Design an incrementality or holdout test for a channel
A test design that measures whether a channel actually creates extra sales, not just credit for sales that would happen anyway.
The prompt
Help me design an incrementality test to find out whether a marketing channel truly drives extra results.
Channel or tactic to test: [e.g. retargeting, branded search, Meta prospecting, direct mail]
Why I'm questioning it: [e.g. "platform reports a 9x ROAS, which seems too good"]
Spend and volume: [monthly spend, conversions, average order or deal value]
Geography and audience: [where we sell, audience sizes]
Tools available: [platform conversion lift studies, geo controls, CRM data]
How long I can run it and what I can risk: [e.g. "4 weeks, can pause 25% of regions"]
Design:
1. Pick the method that fits: platform lift study, audience holdout, geo-based test (matched markets, switch on/off), or time-based on/off. Explain the trade-offs for my case.
2. Define the treatment and control precisely, how to match them, and the primary metric (measured in my own system, not the ad platform).
3. Estimate whether my volume can detect a meaningful lift in the time available, and what to change if not (longer, bigger holdout, bigger spend change).
4. List contamination risks (other channels shifting, promotions, seasonality) and how to control them.
5. Show how to calculate incremental conversions, incremental ROAS or CPA, and the confidence range.
6. Decision rules agreed before launch: what result means scale, keep, or cut.
Return a one-page test plan and a results template. Remind me to verify sample-size math with an analyst if stakes are high.
Highlighted [brackets] are blanks to replace with your own details.
Fill in the blanks and use this promptMore Data & Analysis prompts
What to fill in
- e.g. "platform reports a 9x ROAS, which seems too good"
- monthly spend, conversions, average order or deal value
- where we sell, audience sizes
- platform conversion lift studies, geo controls, CRM data
- e.g. "4 weeks, can pause 25% of regions"
How this prompt is built
This prompt follows the Few-Shot structure (Few-shot prompting). A few-shot prompt includes one or more worked examples of the input and the output you want, so the model can copy the pattern instead of guessing it.
Related prompts
- Design an A/B testHypothesis, variants, metrics, sample size, and duration for a valid test.
- Break down my results into the business drivers behind themBuilds a driver tree from your results, sizes how much each driver contributed to the change, and names which ones to act on.
- Identify the leading and lagging indicators for my business or teamSorts your metrics into leading and lagging indicators, links each lead to the outcome it predicts, and fills gaps in early-warning measures.
- Find the 20% of activities driving 80% of my resultsRuns a Pareto analysis to show the few activities, customers, or products producing most of your results, and what to cut or double down on.