How do we un-fu***k digital advertising?

How do we un-fu***k digital advertising?

To say that digital advertising has had an animated last two weeks would be an understatement. Last week, the ANA "discovered" that advertisers had been ripped off to the tune of 91% -- 21% of the 23% of programmatic impressions that are non-mobile went to Made for Adfraud ("MFA") sites. This week, an outside researcher published 200 pages of damning evidence, collected since 2020, showing 80% of YouTube ads sold by Google as "TrueView, skippable, in-stream video ads" were in fact misrepresented out-stream ads, mostly running on crappy Google Video Partner sites (not on YouTube) and muted, auto-play, and off-page.

The question I have been getting most in the last two weeks is "what do we do? how do we fix this? -- "how do we un-f**k digital marketing," as it were. What may seem to be a scary-monumental task is actually supremely easy. And the answer has been staring us in the face for at least the last 10 years. Ponder for a minute how seemingly impossible that may sound, before I reveal the answer below.

(screen shot of my recent LinkedIn articles below is to create the "pregnant pause" before you get to the answer further below. scroll down when ready)

No alt text provided for this image

How do we un-f**k this? Easy. Get rid of the "easy button" when doing digital marketing.

For example, stop using PMax ("performance max") where you let Google spend your money. When you use PMax, you can't opt out of Google Video Partners ("GVP"), so the TrueView ads you paid for, and thought were running on YouTube, were running muted, auto-played, and off-page on crappy sites outside of YouTube. Do you think PMax maximizes your performance? or maximizes Google's profits? Yeah, it's the latter. I'm not even saying that Google is doing this in a deliberate, nefarious way. It's just that the signal that the PMax algorithms use is primarily clicks.

Having researched ad fraud for 11 years, I can tell you that fraud sites ALWAYS have higher clicks than real sites, because the bots that load the ads also click on them. Video ads that run on GVP have higher clicks (and lower "skips") than ads that ran on YouTube; so the PMax algorithms faithfully allocate more and more of your budgets to GVP, wasting more and more of your dollars in the process. Google makes revenue either way. So as long as they can say "look, we are maximizing your performance" and you believe that more clicks means more "performance," the game of musical chairs continues and no one's the wiser.

Let me add one more point. I empathize with advertisers, especially brand marketers that are shooting for reach and frequency. Clicks are not the main goal; brand awareness and lift are. Many brand advertisers are not measuring brand lift from their digital campaigns (too difficult). SOME are measuring brand lift, but the methodology and results take 6 - 12 months, far far too long to impact digital campaigns that are running. That is why brand marketers are forced to use measurable things like clicks for their digital campaigns and are forced to rely on reporting by the adtech vendors, essentially no different than "grading their own homework." These are true, but ... you can do things differently going forward.

Instead of using the "easy button" and having the black box algorithms of PMax spend your money, select the YouTube channels that are most pertinent to your brand and vertical and create an inclusion list of those channels. Create a new campaign line with this approach. Note your media agencies will push back hard against this (hint: because they make less revenue when you are more informed and take matters into your own hands). Also, when you set up YouTube campaigns, be sure to select the radio button "Only on YouTube" so your ads run as proper "TrueView, skippable, in-stream video ads." There are successful YouTube campaigns where most of the ads were valid "TrueView, skippable, in-stream video ads" and most of the ads ran on YouTube, where the humans are, instead of video partner sites that no one has ever heard of.

No alt text provided for this image

Years ago when I documented fake sites loading hundreds of YouTube embeds on webpages devoid of content, the simple recommendation to advertisers is to turn off ads in embedded YouTube videos, to categorically avoid this form of fraud. As far back as 2014, other academic researchers documented this form of fraud: https://www.researchgate.net/publication/280630756_Understanding_the_detection_of_fake_view_fraud_in_Video_Content_Portals The graph below, from the research paper, showed that "monetized views" were 10 times larger than the "view count" displayed on YouTube. How is that even possible? When you understand ad fraud, you easily understand how bad guys gamed the system to got paid. Large advertisers seeking large quantities of ad impressions got what they wanted AND got ripped off at the same time. Large numbers of YouTube ad impressions reported in dashboards and spreadsheets doesn't mean anyone actually saw your ad.

No alt text provided for this image

Let me not belabor this any more. I think you get the point. When you run Facebook ads, be sure to turn OFF Facebook Audience Network (screen shot below) so your ads remain on Facebook, Instagram, and WhatsApp (which humans use, and are logged into all day long). By turning off FAN you categorically avoid all of the fraud that pervades those sites and apps outside of Facebook. All of them have the motive (to make more money) and the means (bot traffic and other fraud techniques) to siphon your budgets into their own pockets and make you believe you were doing digital marketing. It was literally "make believe" as it were. ;-)

No alt text provided for this image

When you run TikTok campaigns, be sure to turn off Pangle, their audience network. See the 4 pairs of donut charts below to see the difference.

No alt text provided for this image

If you don't turn off the audience network, you will get massive quantities of impressions and clicks, but virtually none of either will be on TikTok where the humans are. See below. All of the clickthroughs with "utm_source=tiktok" came from crappy, fraudulent apps listed below. Turn off the audience network.

No alt text provided for this image

Finally, if you are impressed by how many clicks you are getting from programmatic channels and think that those clicks equate to "performance" think again. You didn't have this data before because legacy fraud verification vendors never showed you this. In the 15 examples below of clicks coming from utm_source=programmatic, the vast majority of the clicks were from bots (orange and red) and only 1 - 8% of the clicks were from humans (dark blue slivers). You got lots of clicks from programmatic, but that was not because the targeting was working well, or your ad creatives were working well. It was from the bots that caused the ad to load, that also clicked on the ads to trick you.

No alt text provided for this image

Humans "google" things. So if you limit your ads to google dot com, those work well. Humans are logged into Facebook and Instagram all day long. So if you limit your ads to Facebook and Instagram and turn off all the sites outside of that, your ads will work well. Humans visit real publishers' sites that they've heard of. If you create inclusion lists of these sites, you will get your ads in front of humans (see examples in the slide below).

No alt text provided for this image

What you won't get are the massive quantities of ad impressions to buy and the large number of clicks. Obviously you realize that most of those ads are not shown to humans or seen by humans anyway, so they won't produce any marketing outcomes for you. You will also see an increase in CPM prices. Don't worry, though, because your total costs may still come down. How? Even if you pay higher CPM prices, if you buy less quantity you will still save money.

All of you are probably thinking the above is "easier said than done." I understand. When I show an advertiser that 90% of their impressions went to fraudulent site and apps, what do you think I tell them? I tell them "don't worry." We are NOT going to chop off 90% of your volume overnight. What we will do is progressively clean the campaigns, week after week, and month after month. We do this by identifying the "worst offenders" -- the top 10 to 20 bad sites and apps and add those to the block list. Then a week later or a month later, we add the next 10 - 20 bad sites and apps to the block list. This way, we prevent more and more of the budget from going to fraudsters so more and more of the dollars can do to better sites and apps. And you need to increase your CPM bids along the way too. This process may take 6 months. It may take 12 months.

As you can see, the key to un-f**king digital marketing is to have a transition plan. When your boss asks you next Wednesday (the day after the July 4th long weekend in the U.S.) about MFA sites and the YouTube scandal, you can tell them you are looking into it, and you have a plan for go-forward. There should be no "FOFO" (fear of finding out) because when you "find out" you can do something about it. Previously, when you get reports from legacy fraud vendors that say IVT/fraud is 1% you didn't have sufficient details to take action. You are welcome to use FouAnalytics to get more detailed "analytics for digital media." Armed with better data, you can do better digital.

Stopping the use of the "easy button" doesn't mean it's complicated and you will be deluged with work. In fact, you'll find it refreshingly easy when you have the right data to make the right decisions, with respect to your digital media. The vendors that sell your services and that grade their own homework are all going to tell you everything's fine, keep spending; they give you as little useful information as possible to keep you in the dark so you will keep spending. After 10 years of "90%" scandals and debacles, hopefully this latest one will wake advertisers up to the fact that no one else will un-f**k digital marketing for you.

By having your own tools -- for example FouAnalytics -- to monitor and manage your digital campaigns you will have newfound freedom, insights and control. Don't let all of the news from the last 2 weeks blow over by the next news cycle, as it has for the last 10 years. Take advantage of the current window of opportunity to take action (before your CEO or CFO ask you about it next week).

"Only YOU can prevent forest fires." (Smokey Bear)

"Only YOU can un-f**k digital marketing." (Dr. Fou)

Let me know if I can be helpful. Happy 4th of July weekend, y'all!







Read more: https://www.linkedin.com/today/author/augustinefou

Watch more: https://www.youtube.com/@augustinefou

Jose Felipe, CMX

Marketing | Trade Marketing | E-commerce | BI | Digital | Advertsing | Development of strategies focused on consumer journey

4mo
Sebastiaan Brouwers

Growth Manager | E-commerce Manager | Marketing Manager

4mo

I love this! Thanks 😁

Like
Reply
Frank Dekker

Expert portfoliomanager bij APG (Telecom and Media investor)

4mo

Ad fraud is an important ESG risk. Meta, Alphabet should prioritize client interest. For their their AI solutions like performance max they should optimize for Roi. Why do believe they optimize for clicks rather than Return on investment? Can customers choose the metric to optimize for?

Like
Reply
Paul Walsh

270 tech & security firms license my mobile app security patents. Patents pending for SMS. Helped to launch AIM in 1997. Co-invented the concept of classifying user accounts & folders on the web in 2004 #dyslexic #ADHD

9mo

Get enough advertisers to care enough about their employers’ money?

Like
Reply
Gregory Wester

Chief Marketing Officer, On-Device Business Unit

10mo

As always, FOUndational analysis!

Like
Reply

To view or add a comment, sign in

Insights from the community

Others also viewed

Explore topics