The short answer
Advertising spend can appear before all associated conversions do. A person may buy later, or reporting may arrive after the transaction. Recent ROAS can therefore be incomplete. Google’s conversion-lag documentation describes delay information for metrics including CPA and ROAS.
Track when a result was observed
Keep the campaign dates and the report extraction date. If the platform credits later conversions back to an earlier interaction period, the earlier period’s revenue can change. If it reports by conversion date, the pattern may appear elsewhere. Confirm the date basis rather than assuming every report behaves identically.
A result can mature without more spend
Illustrative inputs: spend for a fixed acquisition period is $1,000. An early report shows $1,500 of credited revenue, so ROAS is $1,500 ÷ $1,000 = 1.5×. A later report for the same period includes $1,000 more credited revenue. The revised ratio is ($1,500 + $1,000) ÷ $1,000 = 2.5×. This illustrates a possible reporting pattern, not a guaranteed uplift for recent campaigns.
Do not invent a universal waiting period
The useful observation horizon depends on the purchase cycle, attribution rules, and reporting process. Study your own distribution of time to conversion using the platform’s available reports and order records. A fast repeat-purchase offer and a considered business purchase may have different delays.
Compare periods at a similar maturity
Comparing a newly ended week with a much older fully observed week can bias the recent period downward. One approach is to keep age-matched snapshots, such as each period observed after the same elapsed time. Choose that elapsed time based on the business and document it. Another is to label the recent period provisional and revisit it later.
Keep forecasts separate from actuals
If a platform or internal model estimates conversions still to come, show the estimate separately from observed revenue. Record the estimation method and uncertainty. Avoid increasing the observed numerator with an unexplained multiplier. Refunds can also lower later revenue, so maturation does not guarantee that every result improves.
Use early indicators to detect obvious tracking or delivery problems, but make longer-term performance decisions with an understanding of how much of the revenue is still unobserved.
Track report maturity
- Gather: Save the acquisition dates, extraction dates and observed revenue from successive snapshots.
- Produce: Compare cohorts at the same elapsed age or label the newest result provisional.
Decision boundary: Revenue may rise from later conversions or fall from adjustments; neither is guaranteed.
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.