This week at AUSAM we are celebrating our first guest post! While viewability has been a helpful tool to talk about inventory quality and engagement and brand KPIs; the fervor around this topic has exaggerated as buyers look to “100% viewabilily” to solve their digital woes. This week our good friend – and technical black belt – Nadalie Dias takes a deeper look at viewability and why we shouldn’t get too caught up on that 100% goal.
What is viewability?
The first thing we should get straight is actually defining viewability. It does not mean that your ad was seen, was above the fold, on a premium site, or that it was not affected by fraud or non-human traffic. The MRC defines a viewable impression as a display ad in which 50% or more of its pixels appear on-screen for at least one continuous second, or a video ad where 50% of its pixels appear on-screen for at least 2 consecutive seconds. Sounds legit, right? Who doesn’t want to eliminate media waste and by ensuring someone “sees” their ad. But is it as good as it seems?
Aside from the fact that viewable doesn’t mean someone actually saw your ad – it simply means half of your ad appeared on screen for 1 or 2 continuous seconds and, theoretically, had the “ability to be seen” – let’s look at how this is measured.
How is it measured?
Over the years there are businesses that have cropped up whose bread and butter is measuring viewability. And while measuring viewability could be it’s own separate post, in its simplest explanation “viewability” is actually a viewability rate, which is the number of viewable impressions over the number of measurable impressions. How can you measure an impression to determine if it’s viewable? As much as we all think that advertisers are tracking your eye movements and the elements on the publisher’s page when their ad is displayed, the technology isn’t that sophisticated. Here is how a verification provider will determine if an impression is viewable. They look at the following data points: size of the browser window (also known as a viewport), where on the page an ad appears and the size of the ad. Seems simple you look at size of the browser, where the ad is on the page and the size of the ad, right? Unfortunately this method doesn’t work in every instance, and also doesn’t work for cross-domain (AKA “hostile”) iframes or in webkit browsers.
Cross-domain iframes are used by developers as a way to reference content from another website (domain) within their own site (which has a separate domain). How is this used in display advertising? Well publishers’ use this cross-domain iframes in advertising to have a designated ad space on their site where advertisers’ banners/ videos can appear, but the advertisers don’t have access to publisher code on the page. Real talk- it’s a security measure to prevent anyone making unauthorized changes to the publishers’ site through the ad tag java script code. A webkit is a web browser language, it’s used to create browsers such as Safari and Chrome, and it is also used to create apps as well.
That being said, a lot of the time viewabiiity isn’t available, so verification partners will use sophisticated models to create “predictive” scores that use ad serving data to give a best guess on whether or not they think a site and ad format are viewable. Any impression that can’t be modeled will be considered unmeasurable. Altogether, most verification providers can measure, or provide a predictive score for, anywhere from 85-90% of all impressions served.
Now that we understand how viewability is defined and measured, let’s discuss why 100% viewability may not be obtainable - or even worthwhile. Since the viewability numbers you are getting are actually a ratio, you will never be at a point where every impression is 100% measurable in an open exchange RTB environment. Even if measurement wasn’t a problem there is another thing to consider – cost. Advertisers’ may like the idea of viewable inventory, but Publishers’ really love it. Viewability allows publishers’ to charge a lot more for much of the same inventory you were already buying. The difference is that once they can affirm that inventory is viewable with a third party, the CPMs skyrocket when sold in a “highly viewable” or “xx% viewability” package.
For most advertisers the much higher cost of 100% viewable inventory outweighs the benefits. Additionally, it’s been shown that increased viewabilty isn’t directly correlated to return. For the most part increasing viewability does help improve performance up to a point, but there isn’t a large shift in performance between 70-100% viewability.
Which is likely why the IAB guidance issued 70% viewability as the ideal threshold. Because that is where returns in performance start to diminish.
Viewability can certainly be a good thing for both buyer and seller, but make sure you are asking the right questions of your vendors and using appropriate benchmarks for your needs. Many companies have made this issue needlessly confusing with varying definitions, measuring methodologies, and reporting discrepancies. Pop quiz: how many of you realized that the most popular viewability vendor, MOAT, uses 30% of pixels loaded for some ad formats and doesn’t follow MRC guidance on their poll rate? Not saying this is bad – just saying you should know.