PromptSharp › Prompt Library › Marketing › Incrementality test you can actually run and actually read

Incrementality test you can actually run and actually read

Leadership wants proof the channel drives incremental revenue. Design the holdout your budget, timeline, and traffic can genuinely support.

The prompt — copy and run it

You are an experimentation designer scoping an incrementality test. I will paste the channel context and constraints. Produce:

A) THE DESIGN: the test type suited to my constraints (geo split, audience holdout, or on/off flighting), the treatment and control definitions, the duration, and the budget it consumes — with the reasoning for each choice tied to my numbers.
B) THE READABILITY CHECK: given my baseline volume and variability, the smallest effect this test could detect, and the verdict — RUNNABLE, RUNNABLE ONLY IF (state the change to duration, budget, or scope), or NOT RUNNABLE AT THIS VOLUME. Show the arithmetic. Never call a test readable to be encouraging.
C) THE PRE-REGISTRATION: the exact metric, the decision rule written before launch, the contamination risks specific to my setup, and the action we commit to for every possible result — including the result where nothing is detectable.

My context: [PASTE: channel, current spend and conversions, period-to-period baseline variability, geography or audience structure, budget available for the test, timeline, measurement stack]

Rules: Do not invent, estimate, or fabricate any statistic, benchmark, or performance figure — if a number is not in the material I give you, write "not provided" and flag it. Mark every claim I should verify in my analytics or source systems before it is published or presented. Never include customer personally identifiable information or client-confidential terms.

Why this prompt works

Most incrementality tests run at volumes that could never have detected the effect being argued about, and the flat result then gets read as proof of nothing rather than proof the test was too small. Putting the minimum detectable effect BEFORE launch, and pre-registering the decision rule, is what makes the result binding instead of renegotiated after the fact by whoever dislikes it.

Frequently asked

Related Marketing prompts