As the tension grows between publishers and advertisers around viewability, fraud and overall media quality, it’s clear that both sides’ interests are not completely aligned. Both publishers and advertisers have armed themselves with viewability data for their own needs, which are often conflicting. Advertisers are pushing to tie these viewability metrics to payment terms, effectively making viewability a new currency.
Understandably, advertisers don’t want to spend another dollar on ads targeted to fraudulent bots or that have no chance to be seen by humans. Simultaneously, publishers are looking for any data that can prove they have high viewability and no fraudulent traffic. On the surface, these two goals do not appear that far apart, but when you start to dig deeper, you see a different picture.
The first challenge is there are no consistent results regarding viewability among the different vendors in the market. Several factors are involved including difference in methodologies, coverage, use of technology and where in the ad call chain the technology sits. The differentiation starts with the definition of a viewed ad impression. From a publisher’s perspective, an ad can be viewed as soon as its first-party ad server calls the advertisers’ third-party ad server. From an advertiser’s point of view, an ad can be viewed once its third-party ad server sends the ad creative and it is rendered on the website. One key to understanding the difference is comparing the total number of measured impressions to the third-party ad server impression counts. Third-party viewability measurement counts should match up closely to the ad server impression counts. Publisher-focused viewability solutions will often not.
Why is this difference important? Well consider the common scenario where a person goes to a website from a search engine, stays for four seconds as the content loads, and then goes back to the search engine because the website wasn’t relevant. Typically, the ad is displayed several seconds after the page is loaded due to the fact that ad creative comes from the third-party ad server, which is not near the publisher’s content server. The publisher’s viewability clock starts at the page load – and ad call – while the marketer’s clock starts several seconds later – when the ad reaches the publisher page. In a world where one second in view constitutes a viewed impression for display, every second makes a significant difference.
Another challenge is the detection of fraud. As the scientists at Integral’s fraud detection lab say, in fraud there is no ground truth. No one knows how much fraud truly exists, making it difficult to determine whose solution detects the most fraud based on fraud rates alone. This fact allows any vendor with viewability technology to seemingly stand on equal footing when it comes to detecting fraud, regardless of how robust their technology. From an advertiser’s perspective, this fraud detection technology needs to be best-in-class as every dollar spent on a bot is a dollar that is completely wasted. For a publisher, the same financial incentives do not exist to invest in the most robust fraud detection.
Finally, this misalignment of interests does not end at fraud and viewability. Publisher-focused viewability solutions provide additional data points to help publishers rationalize higher engagement and thus higher CPMs. One example of this is mouse hover rates over an ad, or “dwell time.” At Integral, we have collected and observed this data for the past four years, but stopped presenting it to marketers when we saw that it had no connection to engagement or brand lift.
In a comprehensive study, our data science team compared the hover rates of millions of ads that were replaced by blank space to hover rates on actual ads on the same campaigns. Not only were the hover rates over the blank spaces not statistically different than the hover rates over actual ads, the hover rates were slightly higher for the blank spaces.
Interestingly, our data science team also saw a statistically significant link between fraud and hover time. When a marketer focuses on hover rate, the best case is the data will provide no benefit and the worst case is it will lead to higher fraud. This data is certainly not a justification for higher CPMs in the eyes of a marketer, yet the data is actively provided by publishers to justify engagement. This is not to say that quality publishers don’t have higher reader engagement rate, they certainly do, but stats like hover rate don’t depict this and are useless to marketers.
Viewability should be an objective metric reflecting the chance for users to see ads. Marketers and publishers are looking at it from two perspectives facilitated by vendors who serve different masters. Publisher-based viewability solutions favor publishers, while marketer-focused solutions look at viewability and media quality from the lens of a marketer. I expect the industry will eventually come to an agreement to bridge the gap, but until then, there are going to be a lot of negotiations based on different data and view points. Whether you are a marketer or a publisher, understanding the differences in approach to viewability and fraud and understanding the metrics that truly demonstrate better engagement is a good starting point.
Read the full article here.