What does the ROAS Audit measure?

The audit subtracts platform-disclosed inflation sources from your reported ROAS to estimate a debiased range. In connected mode it pulls real account data and compares conversions across attribution windows; in quick-estimate mode it applies industry-research per-platform ranges to your reported ROAS. The result is a corrected ROAS range, not a single point estimate, with each inflation source flagged as measured-from-account or industry-estimate.

What is the difference between connected and quick-estimate mode?

Connected mode requires Meta or Google Ads OAuth and reads real conversions across attribution windows, classifies campaigns as prospecting vs retargeting, and measures inflation directly from your account behavior. Quick-estimate mode runs without OAuth and applies Cresva's industry-research per-platform inflation ranges (documented in the methodology block on the tool page) to your reported ROAS to produce a directional debiased range.

Where do the per-platform inflation ranges come from?

Cresva industry estimates compiled from public attribution research and incrementality studies. Specific ranges and citations live in the methodology block on the tool page. The estimates are directional approximations for the quick-estimate mode and are clearly labeled as such; connected mode uses these ranges only as fallbacks for inflation sources that cannot be measured directly from the account (e.g., cross-device duplication).

Is the corrected ROAS a forecast?

No. The corrected ROAS describes historical attribution behavior, not future performance. It is the same period's reported ROAS adjusted for measured and estimated inflation sources. Confidence intervals widen as the estimated portion grows. True incrementality requires controlled holdout testing the tool does not run.

Does the tool estimate ad spend or revenue?

Spend and revenue inputs are either pulled from your connected ad account (connected mode) or supplied by you (quick-estimate mode). The tool does not estimate or invent spend or revenue figures. Phantom revenue is reported revenue minus corrected revenue, computed deterministically from the inflation range.

How is severity classified?

Severity is computed deterministically from the corrected-ROAS midpoint and inflation midpoint: critical when corrected ROAS is below 1.5x or inflation exceeds 50%; warning when corrected ROAS is below 2.0x or inflation exceeds 40%; healthy when corrected ROAS exceeds 2.5x and inflation is below 30%; moderate otherwise. Parker's narrative summary is produced over these computed numbers but does not pick severity itself.

Your catalogue, open to AI shoppers.

See how

How wide is your attribution-window inflation?

Parker debiases reported ROAS by subtracting platform-disclosed inflation sources.

ParkerAttribution