Start with the question
Write one sentence: “This week, we need to understand ______ so we can decide ______.” A useful question is narrower than “show me competitors’ ads.” For example, investigate whether your next batch should explain a use case or lead with an offer.
Declare the research scope
Record the advertiser Pages, selected markets, requested dates, actual available observations, and media formats included. State which checks are partial or old. Avoid comparing raw totals from unequal samples without an explicit caveat.
Record three observations
For each observation, write what the ad actually shows and attach the record reference. Keep statements concrete: the headline mentions a bundle, the visual shows a single product, or the offer includes a stated condition.
Write a competing explanation
Ask what else could explain the pattern. Additional records may be variants. A change may be a catch-up collection. A missing result may be a failed check. This prevents a neat story from becoming stronger than the evidence.
Select a small test set
Choose one to three ideas relevant to your product. State the customer problem, original message, necessary proof, visual direction, and single variable being tested. Exclude concepts that require unsupported claims or someone else’s creative assets.
Define your own evaluation
Decide where and how you will measure the test using your own systems. Do not use a competitor’s observation count as your success criterion.
Save a concise conclusion
“We observed ____. The uncertainty is ____. We will test ____ against ____. We will evaluate it using ____.” Keep the supporting records accessible so the next review can revisit the reasoning.