Prostujemy: fakty o walce z oszukańczymi reklamami w Polsce
Meta clarifies that its automated ad removal systems utilize a broad spectrum of policy enforcement criteria beyond fraud detection. The company explicitly rejects the methodology used in the Instrat report, citing the misuse of EU-wide reach data to estimate Polish-specific ad revenue.
Verified State Diff
Impact & Verification Analysis
Policy researchers, advertisers, and Polish users of Meta platforms.
This highlights the difficulty in auditing black-box ad moderation systems and the discrepancy between external data scraping methodologies and internal platform enforcement metrics.
Full Fact Overview
Meta's response serves as a technical rebuttal to an external audit regarding their ad moderation pipeline. The core of the dispute lies in the methodology of the Instrat report, which Meta claims conflates general policy enforcement (IP violations, restricted goods, quality standards) with specific fraud detection. Meta highlights that the report's revenue estimation model relies on extrapolated EU-wide reach metrics rather than localized Polish data, and utilizes an arbitrary CPM (Cost Per Mille) of 20.65 PLN. Furthermore, Meta identifies a 'sampling bias' in the report, noting that the data was derived from a single iOS device specifically trained via click-behavior to attract fraudulent content, rendering the findings statistically non-representative of the broader user base.