Checking Viewability Measurements with FouAnalytics
Everyone wants viewable ads, right? After all, if the ads didn't have "the opportunity to be seen" (i.e. "viewability"), they'd have no impact on marketing outcomes. Makes sense. Furthermore, everyone's been measuring viewability for years with vendors like Moat, Integral Ad Science, and Doubleverify, and viewability is almost an afterthought now. Everyone just assumes it's no longer an issue.
But, should you assume the measurement of viewability by these vendors is correct? Given everything else that I have seen, from the failure of detection of bots and fraud to the absurdly incorrect brand safety labeling and blockin, I would suggest you don't just blindly trust their numbers. Why? For starters, remember the case where Newsweek altered the viewability measurements so all their ads were marked viewable? Fraudsters continue to falsify viewability measurements today, so all their ads are marked viewable. Have these viewability measurement vendors corrected for this? Next, did you know fraudsters and bot makers actively strip out the verification tags from these vendors? I've observed code samples dating back to 2014 where bots were deliberately blocking the verification tags from these vendors so they'd have no measurement, and actively tricking the detection by passing false data -- e.g. "not a bot, viewability=1, status 200 OK." Have these detection vendors corrected for these or at the very least report what portion was not measurable? Are they extrapolating measurements from sampled data? Not sure about any of these things. Why? Because it's all black box. They don't and won't explain how they got the measurements. So be sure to press them hard for explanations of how they got their viewability numbers, IVT numbers, etc.
Existing viewability measurements
According to IAS's H1 2022 Media Quality Report, published a few days ago, they report viewability in desktop display to be 73.4%, viewability in mobile web display to be 70.1%, and viewability in mobile app display to be 87.7%. Yay! Our job is done, right?
Not so fast. How did they get those numbers? And are those numbers correct? Let's assume I give them the benefit of the doubt for the time being. After all, they've been doing it for years. But we're still not sure if they are doing it correctly.
So, let's check their numbers with FouAnalytics.
FouAnalytics measurement of viewability
FouAnalytics uses javascript to collect data, just like Google Analytics, Adobe Analytics, and these other detection vendors. For viewability measurement, we focus on the following javascript parameters: 1) visibilityState -- the browser tells us if the tab is active and if the browser itself is minimized. This is an ingredient in the viewability measurement. The tab should be the active one and the browser not minimized for the ad to be viewable, by humans. 2) intersection -- the browser tells us what percentage of the pixels of an ad slot (e.g. 300x250) intersections with the viewport of the browser (the viewable portion inside the address bar and scroll bars). This is another ingredient in the viewability measurement. The value of this must be 50% or greater, since the MRC standard for display ad viewability says "50% of the pixels of the ad in view for 1 second." The "in view for 1 second" is very hard to achieve, especially in mobile environments. This is because humans continue to scroll down the page AND the ad takes too long to arrive in the mobile device and render on screen due to limited mobile bandwidth. Be sure to ask your viewability vendor to prove to you they are measuring this accurately, by showing you the data. 3) hasFocus - the browser tells us if it is the application that is in-use. A human cannot see an ad in the browser if they are using Microsoft Powerpoint and the browser is not the application that "has focus." And finally, 4) unload (milliseconds) -- this is when the user left the page. If the person left the page in under 1,000 milliseconds, the ad certainly could not have been "in view for 1 second (1,000 milliseconds). In FouAnalytics, we report both "viewable" and "viewable-MRC." The difference is simply that the latter includes the 1 second requirement. Viewable-MRC should be a subset of "viewable" because only a portion of the ads that start as "viewable" will also meet the 1 second requirement. By showing both, you can check if the numbers make sense. Some ads that start off as viewable, but did not meet the 1 second requirement, might still be useful ads because all it takes is a glance -- the person scrolling down a page on their mobile device will still notice the ad, even if the ad technically didn't reach 1 second in the viewport.
The above is how we measure viewability in FouAnalytics. Below is what we are observing in open exchange campaigns.
You will note that across the board, the viewability numbers detected by FouAnalytics is half of what the other vendors are reporting, and have reported for years. So who is right? Well consider the fact that from the above, you now know exactly how we measure viewability in FouAnalytics and each of the 4 ingredient parameters are also listed in the dashboard for you to verify and understand how we arrived at the number. Contrast this to those other vendors NOT telling you what or how they measured viewability.
领英推荐
The following are the supporting data fields. VisibilityState must be "visible." Window size should show that the ad was being measured (e.g. 300x250, 728x90, etc.). Intersection percentage is 50% or greater (50% of pixels intersect with the browser viewport). Unload_ms should be 1,000 milliseconds or greater, to meet the 1 second requirement for the MRC standard. This shows that systemwide, viewable-MRC is 31%.
A quick closing thought, don't optimize your campaigns for higher viewability. Knowing that bad guys actively falsify the current viewability measurements, you may be inadvertently sending more money to the bad guys' fake sites. Those sites always have higher viewability than real publishers' sites, because they cheat and real publishers do not.
Let me know if I can help further with measuring viewability and other things in your digital campaigns.