Agree on the comparison
Record the channel, objective, country, currency, date range and attribution setting. Separate prospecting from returning-customer activity where the available data supports that distinction. Compare both a recent period and a relevant baseline; a promotion or stock delivery can change the interpretation. Keep incomplete campaign coverage visible.
Find the question behind the metric
Fewer purchases do not automatically mean the ad is tired. Check whether the offer, audience, delivery, tracking or landing page changed. Then choose one investigation: creative response, search relevance, product availability or the purchase journey. An explicit question is more useful than a dashboard full of unrelated changes.
Write down the next decision
Create a short review note with evidence, uncertainty, a proposed change and an owner. State the commercial measure and the date for reassessment. Your specialist decides whether the evidence supports a controlled test, a reporting fix or no change. Keep the original hypothesis so the team can learn from the outcome.
A promotion changes the comparison
A fashion team sees lower reported ROAS after a sale ends. Before cutting spend, it separates discounted orders from full-price orders and checks remaining sizes. The review may lead to a narrower product focus. This is an illustrative decision process, not a performance forecast.
Use this at work
- Define a comparable reporting period and campaign scope.
- Choose one investigation and name the decision owner.
- Record the proposed test, commercial measure and review date.
Reference material
Platform guidance checked 3 October 2026. Examples and working checklists are Faccelerate editorial illustrations.