Fraud in video advertising is higher than display with an average of 15% of all impressions from branded clips being obtained via unscrupulous methods.
That’s according to Integral Ad Science’s Q3 2014 Media Quality Report, which states that impressions from display ads delivered by networks and exchanges are 13.7% represented by fraudulent impressions, while direct publishers manage a respectable 3%.
Seeing as fraudulent impressions have little impact on dwell times, video ads managed a viewability rating of 30% – much lower than the rates achieved by publishers and ad exchanges. This took into account time spent with each ad, where networks and exchanges managed a score of 54.3%.
Brand safety
Fraudulent video ad impressions can be gained by hiding the actual URL where the ad appears and using what are known as ‘front sites’ that mimic legitimate publishers.
Impressions are then sold that would not normally be purchased by advertisers, targeting high-value verticals such as automotive, travel, health and fashion.
Earlier this year, digital performance solution company DoubleVerify discovered evidence of a number of fraudulent impressions from digital video advertisements and in April, ad tech agency TubeMogul published a list of websites contaminated with bot traffic, generating phony views on video ads for major advertisers like Nissan and Samsung.
Video takes off
High adoption of video advertising during the last year has earned the platform an inclusion in the Media Quality Report, which updates the industry on four important metrics – TRue Advertising Quality (TRAQ) Score, viewability, brand risk and ad fraud.
TRAQ is an cumulative score based on attributes such as viewability, brand safety and professionalism, ad clutter and ad fraud which provides a comprehensive depiction of media quality.
Overall media quality for video, as measured via TRAQ score, was lower compared to display inventory, with TRAQ score for video reaching 561 compared to display scores of 540 for Networks & Exchanges and 675 for Publishers.