CPG prompt of the day

Retail media readout: platform ROAS vs what you can actually claim

The campaign wrapped and the network's dashboard says the ROAS was great. Write the readout that separates attribution from incrementality.

You are a retail media analyst writing an honest post-campaign readout. I will paste the campaign results — spend, attributed sales, ROAS, new-to-brand share, impressions — plus base sales context where I have it. Produce:

A) A READOUT that separates PLATFORM-ATTRIBUTED results from what is PLAUSIBLY INCREMENTAL, listing exactly which numbers are missing to make an incrementality claim (holdout results, base-sales counterfactual, halo measurement) — and refusing to fill those gaps with assumptions.
B) A VERDICT as a three-row table — what we know (spend, delivery, attributed sales), what we believe with medium confidence, and what we cannot claim yet.
C) THREE specific changes for the next flight — negative keywords, dayparting, item/ASIN focus, or budget shifts — each tied to a number in the results I gave.

Campaign data: [PASTE: spend, attributed sales, ROAS, NTB%, impressions/clicks, flight dates, base sales if available]

Rules: Do not invent, estimate, or extrapolate any figure — if a number is not in the data I give you, write "not provided" and flag it. Mark every claim I should verify against my syndicated data or internal reporting before using it externally. Never include retailer-confidential terms or personally identifiable shopper data.

Why it works — Retail media budgets keep growing while most readouts recycle the network's own attribution as proof. The know/believe/cannot-claim structure is what a CFO actually needs — and naming the missing incrementality inputs builds the case for proper holdout tests next flight.