Commentary

Ad Viewability: Verify It's Human, Not Fraud

by Ari Jacoby , December 16, 2014

Ad viewability was the new thing in 2014. The MRC released new viewability standards earlier this year, requiring 50% of a display ad’s pixels to be in view for a minimum of one second (two seconds for video ads). Although many touted these standards as a huge win, not everyone in the industry finds them sufficient.

GroupM, for example, is demanding more from publishers: It will only pay for ads that are 100% visible. Interestingly, Google recently published this viewability study, revealing that average publisher viewability is 50.2%. And therein lies a giant bid-ask spread.

Publishers frequently argue that viewability is an imperfect metric for advertisers and publishers alike. Viewability can vary among users depending on their behavior on a site, Moreover, there is no single standardized way to measure viewability. For their part, advertisers worry that asynchronous load times between ads and a site’s content make it hard to tell how many ads a user ever saw.

All of the attention on viewability, though, glosses over a key first step: ensuring that an ad’s potential viewer is human. Viewability standards do nothing to address the fact that up to 50% of Web traffic is non-human. An ad can be a creative masterpiece, in exactly the right place on the page of the highest-quality publisher, and yet up to half of its “views” could be coming from bots.

In the absence of verified humanity, viewability is not nearly as important as it might seem. A video posted on CMO Today of bot activity captured by Forensiq shows malware trained to view Web sites in just the way a human user might — scrolling through and lingering on some pages before loading new ones. So while viewability may reduce the problems posed by pop-unders and stacked ads, it is helpless against a sophisticated bot.

Addressing viewability and bot-driven fraud do not have to be mutually exclusive. Properly ensuring that advertisers are reaching real people will address many of the causal issues related to the viewability “bucket.” Understandably, advertisers are focusing on the percentage of pixels in view, but there is more value to be had by prioritizing a human audience.

Advertisers should stop paying for fraud altogether — legitimate publishers won’t stand for it and have gone to great lengths in 2014 to address the concerns. As a next step, verified human audiences need to be built into the advertising ecosystem.

Until humanity becomes an industry standard, advertisers can’t be confident that their advertising budgets are going to genuine human eyeballs. Only companies that have built verified human audience into their product offerings deserve to be able to make fraud-free guarantees.

That seems like a good place to start in 2015.

advertising, commentary, viewability

3 comments about "Ad Viewability: Verify It's Human, Not Fraud".

Check to receive email when comments are posted.

Ed Papazian from Media Dynamics Inc, December 16, 2014 at 8:09 a.m.
I couldn't agree more, Ari. The big questions are how do you measure human viewing and, if this is not possible, how will online ad sellers adjust their CPM pricing to account for "phantom viewers"?
Reply

John Grono from GAP Research, December 16, 2014 at 4:18 p.m.

Ed, it is impossible to 'measure' human viewing, but it is possible to provide reliable robust estimates of it. This can be achieved by using multiple data sources such as verified tagged traffic counts melded with a carefully managed representative longitudinal panel and SDKs that leverage 'big data' third-party sources. In essence you use the verified traffic counts to remove non-human traffic (e.g. black lists), fraudulent traffic (e.g. multiple frames, pop-unders etc) and then ensure that the traffic that gets through that meets the minimum viewability standards. Once you know that quantum you use the panel to establish things like device duplication, longitudinal duplication (e.g. cookie deletion), browser duplication (e.g. multiple browsers or browser tabs open concurrently), other device usage (e.g. browser tab open but not in focus), time-based metrics (e.g. dwell time). Third-party data looks very promising to "fill in the panel gaps" particularly for smaller niche sites whose usage in a panel of even tens of thousands ends up in single digits. Sounds simple eh! I daren't use the h-word though!

Ed Papazian from Media Dynamics Inc, December 16, 2014 at 5:04 p.m.

John, I agree that there are various ways to move closer to "viewing" for online instead of simply assuming it. Even when you use a sensible definition of "viewability"---like 100%---- and take into account all of the electronic indicators that are available---- like holding power, CTRs, mouse movement, etc. you still won't know much about most of the "audience". I think that we will have to work, first, on basics. These include establishing much tighter rules about viewability, ad clutter, positioning, etc. which, at the very least, will bring online ad sellers closer to the real world in terms of value delivered to advertisers. How this will come about is anyone's guess for unlike "legacy media", everyone in digital seems to be going in different directions and there is no concensus on what should be done, let alone, on whose going to enforce it. Some contend that this is the advertisers' problem. I think that's wrongheaded. It's the medium's problem and until much stronger steps are taken by its leading players---acting in concert------there's little hope for a positive resolution.