Fraud in Digital Advertising – ANA/White Ops Report Released!

December 9, 2014

By Bill Duggan

Today ANA and White Ops have publicly released the report, The Bot Baseline: Fraud in Digital Advertising.”  Very exciting!

ANA recruited 36 member companies to participate in this initiative to determine the level of bot fraud occurring across the digital advertising industry. Bot fraud refers to automated entities capable of garnering impressions, clicking on ads, watching videos, and faking viewability, often in pursuit of fraudulent profits.

White Ops tagged the digital creative of participants to determine fraud activity in August and September 2014 covering 181 campaigns. The study measured 5.5 billion impressions in 3 million domains (U.S. only) during its 60 day span. Topline findings:

  • Display ads had an 11 percent weighted average of bots.   
  • Video ads had a 23 percent weighted average of bots. The reason that video advertising has higher fraud than display is most likely because video advertising generally carries premium pricing, and the fraudsters “follow the money.”
  • More specifically:
    • Publishers who purchased “sourced traffic” display inventory from a third party had 52 percent fraud. Traffic sourcing is any method by which publishers acquire more visitors through third parties.
    • Retargeted display ads had fraud of 19 percent. Retargeting is the process of delivering ads to particular users based on their previous online activity.
    • Programmatic display fraud was 17 percent.
    • Impressions coming from older browsers (IE6 and IE7) are about half fraud – 58 percent for IE6 browsers and 46 percent for IE7.
    • Bot fraud varies by time of day with peak activity between midnight and 7am.

Bot activity is most likely under-reported in this study as the study was announced publicly, putting bot traffickers on “high alert," and the August/September time period has lower than average traffic activity.  

More details in our report, which is now public.

You must be logged in to submit a comment.