Stage 1: reduce fraud and problematic delivery
In-ad measurement shows which sites and apps generate problematic patterns. Advertisers can turn that evidence into blocklists, seller reviews and other buying changes.
Ad fraud is broader than bots
Fou stresses that fraud can occur without classic bots, for example when apps load ads in the background while the device is not actively being used. Technically real devices do not automatically mean valuable ad delivery.
Stage 2: optimize toward humans
Once problematic delivery is reduced, Fou recommends inclusion lists of better sites and apps and more deliberate optimization toward signals consistent with real human use.
Inclusion lists still require measurement
Allowed domains can still be spoofed in bid requests. Post-bid measurement therefore remains necessary to verify that the ad actually landed where the buying configuration intended.
Stage 3: compare landing-page attentiveness
More advanced advertisers examine the share of paid visitors who take another action after the click. Fou treats this as a faster proxy on the path to genuine business outcomes.
Maturity means better decision rules
The objective is not an ever larger set of metrics. It is clearer rules about what to exclude, what to prioritize, where to allocate more budget and which uncertainty remains unresolved.
This article describes principles and questions. Specific fraud, bot or quality rates cannot be transferred directly to individual campaigns without understanding methodology, scope and data basis.