The short answer
Two ROAS reports can disagree because they answer different questions. First reconcile revenue scope, dates and attribution. Investigate tracking only after accounting for those differences; a higher number is not automatically more accurate.
Compare definitions side by side
| Check | Record for each report | Possible explanation |
|---|---|---|
| Revenue scope | Campaign credit or total store revenue? | Organic and repeat sales may be present only in the store view. |
| Revenue components | Discounts, shipping, tax, refunds | Different components change the numerator. |
| Date basis | Interaction date or order date; timezone | One sale can appear in different reporting periods. |
| Attribution | Model, interaction types and window | Credit can differ without a missing order. |
| Currency | Original and reporting currency; exchange policy | Already-converted amounts must not be converted again. |
| Maturity | Extraction date and pending conversions | Recent periods can change as outcomes arrive. |
| Events | Order identifiers, values and status | Duplicates, missing events or canceled orders need investigation. |
Do not add overlapping attribution claims
Illustrative example: one $100 store order receives $100 credit in platform A and $100 in platform B. The sum of attributed claims is $200; unique store revenue remains $100. This is a possible overlap, not an assertion that every platform pair duplicates every order.
Use total store revenue over complete media spend for a separately labeled blended efficiency view. Keep campaign attribution for its own question. Neither measure establishes the sales that would have occurred without ads.
Investigate a small traceable sample
- Choose known completed orders you are authorized to inspect.
- Match order identifiers, amounts and currencies to the events.
- Check refunds, cancellations and repeat event dispatches.
- Compare attribution and date rules for orders missing from one view.
- Separate explained differences from unresolved differences.
Do not force agreement by deleting unmatched records. Keep the unresolved amount visible and avoid sending customer information to a public calculator.
Write a useful reconciliation note
A compact note should name both sources, extraction time, initial amounts, explained adjustments and the remaining gap. State which view will support the current decision. The goal is a defensible report, not necessarily identical dashboards.
When to move to a tracking audit
If a known eligible transaction has a missing, duplicate or incorrect event, follow the tracking audit. If events are correct but reports assign different credit, inspect the attribution settings instead. Those paths require different fixes.
Official references: Google Analytics attribution explains credit assignment. Google Ads attribution reports describes report-specific behavior. Check current settings in the account being reviewed.
Next reading
Examples are illustrative, not client results or benchmarks. The calculations use the inputs shown; real-world decisions depend on your records and assumptions. Report a correction.