I tweeted this just now …
.. right after getting off one of the most joyful professional phone calls I’ve had in a long while.
One of the largest national brand advertisers gets it, and has taken steps to optimize their digital media buying, against tidal waves of criticism and pushback from agencies. I’ll keep this brief and cut to the bone (with links if you’re inspired to read more).
Buying Direct Saves >80%
It has been well documented by industry-level studies WFA (2015), ANA (2016), ISBA (2020) that at least 50% of every dollar that a marketer spends goes to ad tech middlemen instead of towards “working media” — i.e. to the publisher for showing the ad.
By buying direct from a mainstream, legitimate publisher, a marketer can save the middlemen fees, also known as the “ad tech tax” so much more of every dollar goes towards showing ads. Also, this simultaneously solves other problems like ad fraud, viewability, and brand safety issues that arise when placing ads across millions of sites that no one has ever heard of, let alone visit in large quantities. Eliminating middlemen and shortening the supply chain is also known as “supply path optimization.” Of course choosing better supply paths means better quality for the marketer; but the single greatest factor for improving performance for the marketer is shortening the path. That’s because every single middleman is maximizing their own profits, while adding dubious value, if at all.
Undisclosed Margins and Principal Trading
It’s one thing for agencies to not have to disclose COGS (cost of goods sold), it’s another thing entirely when agency holding companies engage in principal trading for the purpose of getting obscene profit margins, at the expense of their own clients. Agencies are supposed to act as agents on behalf of their clients, the advertisers. But corporate greed, and the absence of any moral boundaries, has led agency holding companies to “regularly engage in numerous non-transparent practices.” Enough marketers have found this to be the case that the Association of National Advertisers had to step in, in 2015, to commission an investigation, which formally found these practices to be pervasive. Agencies giving kickbacks to themselves using overseas subsidiaries. Agencies taking cash incentives from ad tech companies to foist unnecessary services and procedures on unsuspecting advertisers. The list of grave abuses goes on.
Undisclosed or Unknown Fees Somewhere in the Supply Chain
When buying through programmatic channels, a lot “gets lost in translation” as the saying goes, due to the complexity and opacity of the supply chains. It’s hard to see who sold what to whom and how the ad finally made it to be displayed on the page. Tons have already been written about the obvious non-transparency in programmatic ad tech, when digital was supposed to be the most trackable and measurable form of advertising in history.
More precisely, not just “lost in translation,” money is literally lost. That’s right, the money went missing and could not be accounted for. The ISBA study looked only at the most “well-lit neighborhoods” of programmatic media buying and even then they found that 15% of the dollars could not be accounted for. They euphemistically called this the “unknown delta.” This delta averaged 15% but reached as high as 86% in one case — 86% of the dollars went missing and could not be accounted for. How could buyers and sellers simply not know where the money went? It’s because of undisclosed or unknown fees taken by parties like SSPs, DSPs, etc. in the supply chain. By shortening or eliminating the supply chain (e.g. buying a private deal ID set up by a reputable publisher in an exchange), marketers can eliminate their exposure to these problems. Eliminate, not just reduce.
Fraud Detection Tech Was Like Extortion, Tech Didn’t Work, Wasn’t Actionable
When you buy ads in shady neighborhoods, you need some assurances that the ads were real, just like you’d want to know if the coke were real and not just corn starch, right? So marketers went out and paid for so called “fraud detection” services. Unfortunately, all of these were “black box” which means they tell you a rate of fraud — e.g. 9% IVT (invalid traffic) — but won’t explain how they measured it. Of course they can’t and won’t tell you because they have to protect their secret sauce, otherwise you won’t keep paying for it or the bad guys will find out and work around the detection. But this also means marketers pay for fraud detection but can’t take any action other than asking for a refund after the campaign is over.
But publishers and sellers use their own fraud detection tech to dispute the findings. And because everyone is black box, no one can figure out who is right and whose number is wrong. So the end result is that marketers paid for fraud detection, which may or may not have detected fraud accurately, and after all is said and done, are still left with having to negotiate with sellers on how much money they get refunded since no one knows whose number was correct anyway.
Advertisers, be sure to note how vigorously these fraud detection tech companies lock you into long-term contracts, with air-tight no-out clauses. Makes you go hmmm.
Microtargeting Was Crap, Got Back to Branding
For all the magic that digital advertising was promised to be; reality has been far less magical. Marketers have been led to buy more and more targeting parameters, thinking that would make their ads more relevant to target audiences and lead to better business outcomes. But instead, more targeting parameters has simply led to more business for the ad tech companies selling the data. They make more money the greater number of parameters they convince the marketer to buy.
When more targeting parameters led to audience segments that were too small, the adtech vendors just made up more, euphemistically called look-alike audiences. These audiences were derived from data collected — these users “look like” the desirable audiences you want to buy because they visited the same sites or searched for the same things online. Do you know just how easy it is for bots to visits a specific collection of sites and/or type in specific search terms to make themselves appear to be any desirable audience segment marketers want to buy? It’s easy.
So not only is the data collected highly “sus” (suspect), the data may not have been collected with the knowledge and consent of the users (privacy-invasive, and possibly not compliant with the law), AND bots might be the ones you’re creating “at-scale” audiences from. The data was crap; the targeting was crap; the audiences were not crap, but bots. This marketer went back to digital marketing 101; blocking and tackling, as it were. Buying run-of-site inventory on good publishers’ websites that had real human audiences. At the very least, the ads were shown to humans. Lacking this, digital ads could not have any chance at driving real business outcomes, no matter HOW awesome the vanity metrics looked.
In the 4 minutes since 10:30a ET, I wrote a quick note of thanks “[your hands-on experience going through this process] gives me hope when it seems there are so many storm clouds [in digital marketing] I was going to lose all hope.” That’s why “I’d Give My Ad Budget To This Marketer To Spend For Me In Digital” even though I would never give anyone my money to spend on anything. This marketer’s a good one.