For better digital marketing, do the exact opposite of this agency

Today is April Fool's Day, but this article is no joke. I am dead serious about doing the exact opposite of what this agency holding company says in their press releases.

"Buy 100% viewable"

This agency sent out a press release in 2017 [1], proclaiming they will only buy 100% viewable ads. On the surface, this seems to make sense because if an ad was not viewable, it couldn't work. Ads were "viewable" if 50% of the pixels of the ad were in the viewport of the browser for 1 second, according to the industry standard. While this idea made sense in theory, in reality, it caused more of the media budgets to go to fraudulent sites that were falsifying viewability measurements, and less money to go to legitimate mainstream publishers with real human audiences. It also harmed real publishers whose page layouts meant that 1/4 to 1/3 of their ads were no longer "sellable" because they were not immediately viewable. This agency misspent millions of their clients' budgets on fake sites that made their ads appear to be 100% viewable 100% of the time through fraudulent means.

A "viewable" ad means it has the opportunity to be seen; it doesn't mean anyone actually viewed it.

Doing the opposite of this -- i.e. buying from real publishers even if their viewability was not 100% -- would have yielded better outcomes and better digital marketing.

"Buy 100% fraud-free"

This agency, as part of their association with TAG (Trustworthy Accountability Group) also claimed they would buy "100% fraud free" ads and even provided guarantees to clients that they would refund anything that wasn't fraud free. On the surface, this seems to make sense because who wants fraudulent ads (ads loaded by bots and other fraudulent means)? While this idea made sense in theory, in reality, it caused more of the media budgets to go to fraudulent sites using bot traffic that the fraud detection vendors could not detect. These same legacy vendors repeatedly reported 1% IVT ("invalid traffic") for years. Bots blocked their detection tags so these vendors had no data with which to mark those bots as IVT. So the low fraud numbers were not because fraud was actually low, it was because these vendors failed to detect it. This agency misspent millions of their clients' budgets on fake sites that appeared to be fraud free because they successfully tricked the detection of the verification vendors to appear to be "fraud free."

"1% fraud doesn't mean that's all the fraud there is; the 1% is all the fraud they could detect."

Doing the opposite of this -- i.e. buying from real publishers even if their invalid traffic was not 0% -- would have yielded better outcomes and better digital marketing.

"Block all MFA sites"

Immediately after MFA sites came to light (mid-2023), this agency sent out a press release [2] claiming it was the first to remove all MFA sites from its inclusion lists. On the surface, this seems to make sense because who wants to show ads on sites that were "made for advertising"? All sites that make money via ads are "made for advertising." But all of a sudden, there was a particular list of sites, procured by this agency from a single adtech vendor that were labeled "MFA" (in the nefarious sense of the term). All of a sudden, it was a bad thing to buy ads on these sites, despite the same sites being part of media buys for years, because of the enormous quantities of ads they offered, higher viewability and higher "performance" (i.e. more clicks), and lower CPM prices. But what about certain sites made it MFA? Most advertisers may have a vague understanding of this -- some sites loaded too many ads per page, some used pirated content or AI-generated content, etc. But hadn't fraudulent sites been doing this for the last 15 years already? This agency misspent millions of their clients' budgets on MFA sites over the last 15 years, before these sites were labeled "MFA" by an adtech vendor whose list is being used by this agency to "block all MFA."

All sites carrying ads are "made for advertising." What got some of these onto a vendor's "naughty list" while others didn't? What about the MFA sites that have human audiences?

Doing the opposite of this -- i.e. buying ads from real publishers even if they were accidentally listed as an MFA site by the adtech vendor -- would have yielded better outcomes and better digital marketing. What's more important is whether a site has real human audiences or not, which you can see if you use an in-ad tag from FouAnalytics.

Buy 100% brand safe/brand suitable

This agency tells their clients all their ads are "brand safe" or "brand suitable." But what exactly does that means; or does it mean anything at all? Unfortunately, this holding company agency is using the same legacy verification vendors for fraud detection as for brand safety detection. For years, there have been many documented cases of the brand safety technology blocking ads from legit publishers because the page url or content of the page contained words like "covid" or "blood" or "shooting." If anyone with common sense looks at the page, they will realize that articles about "covid releif funding" or medical journal articles containing the word "blood" or "shooting pool" are all fine for advertisers' ads. On top of that many of these brand safety vendors' pre-crawling of webpages led to completely incorrect categorization of the page -- e.g. a page on winter crafts was categorized as not brand safe due to "s@x with an1mals." (by the way, replacing the "e" with @ and "i" with 1 is exactly how bad guys get around brand safety detection, which is little more than glorified keyword lists.

Buying based on "brand safety" as reported by vendors that can't detect brand safety properly has led to real publishers getting defunded, and more dollars doing to disinformation sites (hint: breitbart still makes money by getting around brand safety detection and blocking).

Doing the opposite of this -- i.e. buying ads from real publishers's sites including news sites, even if the pages contained the words covid, blood, or shooting -- would have yielded better outcomes and better digital marketing.

Buy on "attention"

Finally, this agency group has told all its clients it is buying ads based on "attention." On the surface, this seems to make sense because digital ads only work if humans saw them -- i.e. "paid attention" to it. Using eye-tracking studies in laboratory environments, attention vendors proved that large ads that covered more of the screen got more "attention" than small ads that could be easily ignored. Also, video ads that had higher completion rates meant greater "attention" because users watched more of the ad before skipping or leaving the page. And of course, more "attention" meant higher brand lift metrics like "recall" (humans recall the ad if they paid attention to it, duh). While this idea made sense in theory, in reality, it amounted to nothing more than optimizing for larger ad units or ads that started off viewable -- e.g. above-the-fold or viewable before the user scrolled down an infinite-scroll page.

"higher attention is better for ad effectiveness, of course, but attention vendors cannot actually detect attention"

Attention vendors cannot detect whether anyone was actually paying attention to the ad, because their javascript detection tags don't have permission to turn on the camera to detect if someone was actually looking at the screen. In other words, attention vendors cannot actually detect attention. Instead, they use "probabilistic models" to predict how much attention ads might get. This is not good enough. Furthermore, every metric these vendors use as input to their probabilistic models can be faked by fraudsters to make their ads appear to achieve higher attention. Fraudsters easily and regularly fake video completions, time on site, hover rate, touch events, in-view time, exposure time, etc. Just like they falsified viewability and blocked IVT detection tags to make their ads appear 100% viewable and 100% fraud-free, fraudsters easily manipulate attention metrics so they can steal more dollars from unsuspecting advertisers whose budgets are being spent by said agency, buying on attention metrics.

Doing the opposite of this -- not buying on attention, which you cannot measure, but instead measuring the attentiveness of the users that arrived on the landing page -- will yield better outcomes and better digital marketing. And it will allow you to buy less of the types of ads the agency is pushing on you.

The graph below shows the humanness (dark blue on the donut chart) and "attentiveness" of the users that arrived on the landing page -- i.e. did they do anything else after they arrived (like click something)? Better ad creatives also lead to more attentive humans, because humans who deliberately click an ad are curious about something and will likely do something else on the landing page, like move the mouse, scroll the page, click something, touch the screen, etc.

See: How to optimize media investments based on attentiveness

Are you going to let a big media agency mis-spend your digital ad budgets? or are you going to do "evidence-based" marketing and look at analytics yourself so you can optimize your own media with basic common sense?

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