What is Conversion Quality?

Conversion Quality measures how much of an account's reported conversion volume is observed reality versus estimation — splitting every conversion into click-through, view-through, cross-device, and modeled components.

Why it matters

Two accounts can both report 500 conversions while one has 480 verified purchases and the other has 300 modeled estimates plus 120 view-throughs. Smart Bidding treats both identically; your P&L does not.

How ClickCatalyst detects it

The modeled conversion percentage shows how much of reported data is estimated rather than observed — training-data pollution for the bidding algorithm. View-through inflation risk flags campaigns where mere impression exposure is being counted as performance.

A deduplication check computes an inflation ratio to detect the same conversion being counted multiple times, and the conversion lag distribution shows how long after a click conversions actually land — context for judging any recent "drop".

modeled_pct = modeled_conversions ÷ total_conversions × 100

components: click-through · view-through · cross-device · modeled

Example

A campaign whose conversions are 40% modeled and 25% view-through is optimizing toward a number that is two-thirds assumption — its "CPA improvement" may be an artifact of counting, not selling.