YouTube TrueView Ads are Awesome, Despite the Scandal of Recent Weeks

Here is a campaign I am running and paying for myself. YouTube campaigns can be very useful and impactful, even for small businesses like mine. The large scandal of the last few weeks was due to large advertisers chasing massive quantities of views and clicks and not getting what they thought they paid for -- TrueView ads. TrueView ads are the ones that users choose NOT to skip. Advertisers only pay for ad views. This is great. But the Adalytics research documented 1,100 large brands over the course of the last 3 years, paying for ads that were out-stream (not related to any content video), muted, off-page, auto-played, and non-skippable -- i.e. ads that did not meet the definition of TrueView. Note that this misrepresentation of ads is different than whether the ads ran on YouTube or off YouTube.

The vast majority of MISREPRESENTED TrueView ads occurred when the ads ran on GVP (Google Video Partners) sites and mobile apps. The vast majority of VALID TrueView ads were delivered when they ran on YouTube itself.

So, just like the recommendation to turn off Facebook Audience Network (FAN) when running Facebook ads, or to turn off search partner network when running Google or Bing search ads, the recommendation is the same here -- turn off video partner sites and apps in order to keep all your ads on YouTube itself. That's where the humans are anyway, so all of the massive quantities of impressions and views elsewhere on sites and apps that no one has ever heard of are just that, useless impressions and views. By turning off video partners, you are not only avoiding most of the fraud, you are also getting mostly valid TrueView ads (running on YouTube itself). But obviously, the quantities will be far smaller, in some cases 90% smaller than what large advertisers are paying for now.

The sites and apps in the video partner network may deliberately use various techniques to cheat and generate as many views as possible so they can maximize their own profits. Google obviously makes a portion of this revenue so they have an incentive to look the other way. Google even hides behind outside verification firms like IAS and DoubleVerify, claiming they provide "independent" verification. But that is a lie, since Google supplies the data to these vendors through Ads Data Hub (ADH) and these vendors only "perform calculations" and "provide reporting on" the data given to them by Google. These vendors did NOT measure anything with their own javascript tag and the reports they supply are NOT "independent" verification. Don't buy into their lie.

Obviously, there are SOME honest publishers in the video partner network -- like nytimes.com, washingtonpost.com, reuters.com -- that run TrueView ads correctly. You can use an inclusion list to try to target these handful of legitimate publishers. But note that Google makes it difficult for large advertisers to turn off GVP, for obvious reasons. Video partners generate such an enormous quantity of ads that Google would lose significant revenue if they allowed advertisers to turn it off or opt-out. But advertisers should realize by now that most of these are invalid video ads sold as if they were TrueView so it is of no value to them, despite the enormous quantities.

Below is my campaign. You don't need large quantities of impressions and views to have an impact for your business. You just need valid views -- views of your ad that users chose NOT to skip. Here's my campaign set up for 1 of the ads. The data is for a little over 1 month -- June 18 - August 5, today.

Below are the overall stats. Note that 14.7% of my ads ran on TV screens (green in the bar chart below). Yay! This is a very very affordable way to run ads on big screens, compared to all of the CTV and fraud in CTV I am seeing elsewhere. Note that 1.5% of the views (ads not skipped) occurred on TV as well, but 0.0% of the clicks came from TV, because no one clicks through to your website from a connected TV. Ahem, all those CTV ad buyers who are seeing tons of clicks to their website. Those are falsified clicks (if you want to understand this more, get in touch privately). 99% of the clicks to my website are from mobile devices. FouAnalytics data (on the clicks that arrived on my site) corroborates this.

Below is the overall performance of the campaign. Note specifically the 430 clicks on the ad, as reported by YouTube. Note the 50,241 views, which is 66.6% of 75,400 impressions. Views are views of the video ad that were not skipped. A 66.6% view rate is very high. I would have expected many more users to skip the ad. The CPV (cost per view) is 1.6 cents. 431 clicks out of 75.4k impressions is a 0.57% CTR (click through rate). The clicks-to-views percentage is even better, at 0.86% (431 clicks divided by 50,241 views). The ads were delivered to my target audience ("Advertising and Marketing Services").

And, of course, I don't trust any of the reporting supplied by YouTube so I measure it with my own tools -- FouAnalytics. Since YouTube does not allow me (or any other vendor) to measure the ads themselves, I have my FouAnalytics tag on my own website landing pages so I can see the clicks arriving from this campaign. Obviously you need to use click through urls that have the proper "utm_source=youtube" query strings so you can identify they came from YouTube, and the "video=95_percent_bounce" tells me these came from this exact campaign. Note the quantity -- 429 pageviews in FouAnalytics -- closely matches the 430 clicks reported by YouTube. I would consider this corroborated.

Note the relative quality too -- 23% confirmed dark blue. This means of the users that arrived on the site, we could see further interactions like mouse move, page scroll, clicks, etc. on the page. Only 7% of the clicks were marked as dark red. Anything under 10% dark red can be considered "good enough" considering I have seen 90% of the clicks being dark red. Note the pairs of examples below, where audience networks and search partners are not turned off, compared to when they are turned off.

I hope this concrete example of how I am running a YouTube campaign may serve as inspiration for other advertisers big and small. Don't line Google's pockets with billions of dollars of profit by letting them eat up all your ad budget on INVALID TrueView ads running on GVP, video partner sites and mobile apps. Those are useless, and mostly fraudulent, views anyway, not seen by humans. By the way, ON YouTube, most humans skip the ads, so you don't even need to pay for the impressions that were skipped. You pay only for the small portion that are NOT skipped -- i.e. real TrueView ads. OFF YouTube, however, the data from Adalytics shows that virtually none of the ads are skipped (e.g. because the skip button was deliberately covered by something else). So the vast majority of the INVALID TrueView ads you paid for ran on outside sites and apps. Ponder this for just a minute more. Most ads on YouTube are skipped, so only a tiny portion are paid for as valid TrueView ads. Most ads off YouTube are not skipped because of fraudulent techniques employed by the websites and apps to maximize their profits. Most of the TrueView ads big advertisers paid for are the ones running outside of YouTube and misrepresented as TrueView. This specifically was what Adalytics documented for the last 3 years, was completed missed by other verification vendors, and was not addressed by Google's stupid blog post.

There IS a better way to buy YouTube TrueView ads. YouTube TrueView ads do work if valid TrueView ads are purchased. There IS a better way to buy digital ads via programmatic channels and tools. It's just not the way large advertisers are buying it now. There IS a correct way to buy real CTV ads and avoid the other 99% of CTV ads that are entirely made up out of thin air. If you care about these things, get in touch. You are welcome to use FouAnalytics at no cost, so you can "see Fou yourself" too, like many other advertisers are already doing.

Happy Saturday, y'all! I am BACK!

Take nothing but memories. Leave nothing but footprints.