Facebook ads data is easiest to misread when the review starts with the biggest number on the screen. A rising ROAS, lower cost per message, or more purchases can all be useful signals. None tells the whole story without the period, the business outcome, and the changes that happened alongside it.
Set the comparison before interpreting the result
Write the exact date range, attribution view, account and campaign scope. Then ask whether the comparison period had a similar offer, stock position and spend pattern. This does not make the data perfect; it makes the limits visible to the person making the decision.
Follow the result beyond the ad platform
Match the metric to the objective. A messages campaign may need a check on qualified conversations. A purchase campaign may need an order or revenue signal from the business. If downstream data is absent, record that absence instead of treating a platform metric as a completed business outcome.
Fictional worked example: two campaigns, one period
This invented USD example covers 1–7 August 2026 with the same attribution setting for both campaigns. “Sales” means attributed revenue, not profit. Leads are lead-form submissions; conversations are conversations started, not individual chat bubbles. The two counts may overlap and are not unique people.
| Metric | Campaign A | Campaign B | Total |
|---|---|---|---|
| Spend | $100 | $900 | $1,000 |
| Sales | $400 | $1,800 | $2,200 |
| Lead forms | 3 | 12 | 15 |
| Conversations | 5 | 45 | 50 |
| ROAS | 4.00 | 2.00 | 2.20 |
| Cost / conversation | $20 | $20 | $20 |
Total ROAS is $2,200 ÷ $1,000 = 2.20, not the unweighted campaign average of 3.00. Cost per conversation is total spend ÷ total conversations. Compare only periods, currencies and attribution settings that match. If tracking is missing, write “unavailable”; if a denominator is zero, the ratio is not applicable, not zero. For the ROAS formula only, see Shopify’s ROAS formula.
Read creative and delivery together
Open the ads that spent the money. Consider the message, offer, audience fit, comments and repetition. Also note delivery constraints and any edits made during the period. A creative change and a budget change in the same window are not a clean creative test.
Leave with one test, not a list of guesses
Choose one hypothesis, one change, one owner and one review date. For example: keep the audience and budget steady while testing a clearer opening line, then review the quality of conversations after the next complete period. The example is a process, not a universal performance rule.
Weekly decision note
- Period, comparison and data freshness:
- Objective and verified business signal:
- Spend, sales, lead forms, conversations, ROAS and cost per conversation:
- Creative or delivery context:
- One hypothesis and one change:
- Owner and review date:
Use the full review checklist
Work through six practical steps before you make the next budget, creative or delivery decision.
Open the English checklistExplore CasperAds to keep connected-account reviews and next actions visible to your team.