Dr. Augustine Fou says 90%+ of ad inventory can be fake
Dr. Augustine Fou, an ad fraud measurement expert, explains the complexities of ad fraud within the digital advertising ecosystem, highlighting how fake websites and bot traffic generate fraudulent ad revenue, particularly in programmatic advertising. He details that AI's impact on ad fraud is incremental, primarily aiding in faster content generation for 'made for advertising' (MFA) sites rather than fundamentally altering the nature of fraud. Fou advises marketers to maintain independent measurement tools and not rely solely on platform reports to detect discrepancies and minimize losses to ad fraud.
Key Takeaways
- Fou said bad actors can spin up thousands of WordPress-template sites and use bot traffic to force ads to load.
- He traced the fraud spike to programmatic exchanges, citing AppNexus and a 2016 cleanup from 260 billion impressions to 20 billion.
- Fou said AI is mostly incremental, helping generate recipe sites and other MFA pages faster, not changing the core mechanics.
- He said some five-day-old sites are already selling 10 million impressions, which he called 100% fraud.
- He warned that DSP placement reports can show zero overlap with independent post-bid measurement, because bid requests are only declared data.
Why It Matters
For advertisers, the immediate takeaway is that platform reporting cannot be treated as a single source of truth: Fou described a case where DSP placement reports had zero overlap with independent measurement. The broader ecosystem problem is structural, not just technical — programmatic exchanges, multiple middlemen, and bid requests that are only “declared” create room for fake inventory and toll-taking. The next signal to watch is whether marketers adopt their own post-bid analytics and compare them against DSP placement reports, especially when discrepancies are large enough to change where budget goes.
Read full article at youtube.com