90+ percent completion rates for video ads, are you kidding me?

If you run OLV ("online video") campaigns and the reports you are given are showing 90+% completion rates, you should be skeptical, not accepting. FouAnalytics measures video ads with a javascript tag (VPAID not VAST). The unload_ms data grid shows the number of milliseconds when the page was "unloaded" which means the user left the page. In the example below, 92% of the impressions had unload_ms in the 514 - 17,414 range. That means the user left the page in 0.5 - 17.4 seconds. If your video ad was a 30 second ad, there is no possibility it was viewed to completion. If your video ad was 15 seconds, then this is fine.

The reason you should use independent analytics to check what the vendors' reports say is 1) vendors' reports are a form of "grading their own homework, and 2) they may not be able to tell if fraudsters simply invoked the Complete.gif tracking pixel to make it seem like your video ad was watched to completion, when it was not or wasn't even loaded at all.

Note that in CTV campaigns, since we cannot run a javascript tag, we cannot detect the unload event. This means we have to trust what the sellers/exchanges self-report. The following is one of the only examples where I have ever seen one of the other quartile pixels being called and reported correctly. The VAST video serving standard is further below. Vendors are supposed to invoke the firstQuartile.gif, midpoint.gif, thirdQuartile.gif, and complete.gif tracking pixels at the right time. But when was the last time criminals followed the rules? The rule of thumb in digital advertising is not "trust but verify." It's "never trust, always verify with your own independent tools and common sense."

What portion of your OLV ads were audible?

Most advertisers assume that video ads work better than display because humans remember the ads better if they hear audio. That's true. But, most video ads are not played with sound on. Legitimate publishers don't autoplay video ads with sound on, out of respect for their users. Fraudulent publishers play lots of video ads in the background or in popunders, ALSO without sound on so they remain undetected. So what portion of your video ads have sound on, or are "audible?" Right nearly 0%. I will be generous and round up to 1% of your video ads are audible. From data that I have reviewed, most ads are 0% audible. Only 2 rows show >0% audibility -- cnn.com and Pandora's android app. These make sense because humans do leave cnn.com on with sound on in the background. So video ads get to play with sound on too. And folks listen to Pandora so sound is on when video ads come on too. In all other cases, humans have to unmute deliberately to hear the audio of the video ads. The % in the rightmost column tells you what % of humans do that -- almost none.

So what?

If users leave the page in 0.5 seconds - 17 seconds, then 30 second video ads cannot be played to completion even if the reports you get from platforms show 90% completion rates. Those completion rates are easily falsified because fraudsters can invoke the complete.gif tracking pixel at any time, even if the video ad was not actually played to completion. Video ads are 5 - 10X more expensive than display ads, but most people don't watch them. If only a quarter of users watched video ads through the firstQuartile, and only 1-6% watched through the thirdQuartile, how effective are your video ads at generating awareness? Right. Not very. Better to use display ads for 1/5th the CPM and generate awareness in the first 0.5 seconds. Nearly 0% of your video ads are played with "sound on" anyway so they don't have greater impact than display ads.

With all this new information, please let me know if you decide to pause your OLV campaigns or change them over to display campaigns until they "clear this sh*t up." For some of my clients, they are going back to their OLV "partners" and insisting that FouAnalytics javascript tags be run -- e.g. in VPAID environments -- so all of the following details can be independently measured:

• unload_ms -- when the user left the page, so we can verify if "completes" for 30s video ads are even possible
• what portion of the ad was actually watched -- were the tracking pixels invoked at the right times
• was the sound on (volume level detected by FouAnalytics) and did the video ad actually play

If the platform refuses to allow you to measure video ads with a FouAnalytics javascript tag, you are within your rights to NO buy from them any more, and go with partners, vendors, and platforms that DO allow you to independently verify the video ads you are buying.

Oh, and another no-cost way you can "clear this sh*t up" yourself is by buying video ads from real publishers like Disney, Amazon Prime, Hulu, Netflix, etc. Those publishers are not deliberately trying to rip you off, and those publisher will follow the rules and invoke complete.gif when the video ad is actually viewed to completion. So you don't need FouAnalytics or any other verification anyway if you buy from real, good publishers. You won't be able to detect your way out of trouble if you buy through programmatic channels and your ads end up on unknown sites and mobile apps.

Further reading: https://www.dhirubhai.net/in/augustinefou/recent-activity/newsletter/