There’s been a lot of talk lately about brand safety, owing in part to brands’ ads being displayed in front of extremist videos on YouTube. Programmatic digital ad buyers indeed are prioritizing brand safety to a great degree, according to recent research. So how much riskier is it to buy programmatically than direct? Integral Ad Science (IAS) reviewed its data from H2 2016 to provide some answers in its latest Media Quality Report [download page].
US Display Brand Risk
Based on an analysis of 200 billion impression in the US, IAS determined that 8.6% of all impressions in H2 were flagged on objectionable content, deemed moderate (7.6%), high (0.8%) or very high (0.2%) risk. On an encouraging note, that was down from 9.5% of impressions in H1, though the US media was higher than in any of the 5 other countries highlighted (which ranged from 3.4% – 6.8%).
Notably, though brand risk was more than twice as high for programmatic than publisher direct buys, with almost one-tenth (9.5%) of the former flagged against 4.6% of the latter.
While the types of content flagged was fairly similar for both programmatic and publisher direct, there were some difference. The adult (29.8% share) and offensive language (27.3%) brand risk categories were the most common for publisher direct media buys, both at higher rates than for programmatic buys (26.7% and 19.8% share of flagged content, respectively).
For programmatic buys, violence (29.5%) was the most common offensive content, at a considerably higher rate than for publisher direct buys (21.7%). Hate speech, illegal downloads and illegal drugs were also more common for programmatic than direct buys, though fairly limited for each.
Global Video Brand Risk
Turning to global video ads, IAS’s sampling of more than 6 billion impressions globally found that 8.9% were flagged, with 7.8% deemed of moderate risk, 0.9% of high risk, and 0.2% of very high risk. Again, that rate was an improvement from H1, when 11.2% of impressions were seen as risky.
Once again, the rate was higher for programmatic (9.8%) than publisher direct buys (6.4%), though in this case, the gap was slightly smaller.
For video, violence was the main content category for brand risk (35.5% share of flagged content for programmatic buys and 41.2% for direct buys). As with display, hate speech, illegal downloads, and illegal drugs were more prevalent in the mix of offensive content categories for programmatic buys than direct buys. While these were not as pervasive as violence, hate speech accounted for one-tenth of flagged video impressions that were bought programmatically.
In other highlights from the report:
- Almost one-third (29.4%) of data center traffic was bot traffic;
- The risk of ad fraud increases during the overnight hours, reaching a peak at 4AM;
- Ad fraud risk grows as articles get older; though
- Ad fraud is almost twice as high (88% higher) on home pages than on article pages.
About the Data: IAS notes the following definition of brand risk:
“Brand risk refers to impressions that are flagged on pages that pose various levels of harm to brand image and/or reputation through association, based on seven core content categories: adult, alcohol, hate speech, illegal downloads, illegal drugs, offensive language, and violence.”