The Economics and Incentives of Digital Ad Fraud

This is my 300th article on LinkedIn, and 10th anniversary of FouAnalytics .. so I am reprising some "greatest hits" charts from over the years. Those of you who have been with me all along will recognize them. Hopefully this article will inspire action among others who are new to the topic.

Digital advertising has been around for more than two decades. But the last decade (2012 - present) is what I consider the digital marketing's "lost decade." In an effort to appear "digitally transformed" advertisers large and small have poured their ad budgets into digital, chasing large quantities of ads, low prices, and high "performance." What they didn't know, or refused to admit, was that most of these three things were driven by bot activity and fraud. Waiting for humans to visit sites is not a reliable way to get traffic. But using vast botnets to repeatedly load webpages is super reliable -- you get exactly what you pay for. Botnets create enormous quantities of ad impressions to sell, even on "long tail" sites that no one has ever heard of. This also enables the low CPM prices seen in programmatic channels because fake sites don't have any costs of creating content like real publishers do; and finally the bots are programmed to click on ads so they appear to "perform" better than ads shown to humans. The combination of large scale, low prices, and high clicks is the crack cocaine advertisers have been addicted to for the last decade.

"They've been trippin' for the last decade and not doing real digital marketing ."

Not enough humans on earth

The chart below illustrates this -- the "lost decade" of digital marketing. Since around 2012, you can see the blue line (digital ad spend) diverge more and more from the yellow and green lines (humans' usage of the Internet, social, and mobile). That "gap" has continued to increase to the point where today 2/3 of the digital advertising activity cannot be explained by human activity (interactions with ad supported media). No matter the format of the ad -- display ads, video ads, mobile ads, in-app ads, CTV/OTT ads, etc. -- bots and fraudulent activity have fueled their rise. This created a vicious cycle -- more bot activity, more stuff to buy, more budgets allocated to bot-fueled "digital, which created more incentive for criminals to use even more bot activity, and so on.

The ad fraud crisis started ten years ago, and today in the U.S. alone at nearly $150 billion annually, bot-fueled digital advertising exceeds the illicit drug market -- "drug users in the United States spend on the order of $100 billion annually on all four drugs (cocaine (including crack), heroin, marijuana, methamphetamine (meth). This figure has been stable over the decade." (Source: RAND ). Worldwide, digital ad spending is approaching $450 billion. That means every year, this pot of gold is refilled by advertisers' ad budgets; and fraudsters and criminals worldwide happily gorge themselves on the easy money.

Easy to do, easy to get away with ... and highly profitable

For criminals, digital ad fraud is an easy decision. Why risk your life trying to rob a bank or reselling cocaine when you can sit in front of a computer and do digital ad fraud all day long? Buy traffic for $1 per thousand (clicks) and sell ads for $10 per thousand (ad impressions), netting $9 of pure profit from the comfort of their gaming chairs at home. Who wouldn't put in the next $1 to get the next $9 out. And that is the most simplistic example. There are many other ways they can multiply their profits -- like loading a dozen ad slots on the page, stacking 20 ads in the same ad slot, refreshing the page or the ads every second, etc. For funzzies, the "Criminal Trade-Off Matrix" below shows the large "value of proceeds" from digital ad fraud, the minimal "costs and materials needed," the likelihood of success, and the extent of the risk for doing ad fraud.

If you don't believe how easy this is, see this experiment by Megan Graham, a reporter at CNBC. To show how easy it is for plagiarized news sites to get ad revenue, I made my own . Seven of eight ad exchanges let her site in, despite the domain being less than 1 month old and all of the content being plagiarized (her own articles). Within a week her site started getting ads from the largest advertisers. "I didn’t want to be taking ad revenue from legitimate advertisers, so I only briefly activated advertisements from the partners to see what surfaced and to take a few screenshots. I saw ads come through in for companies including Kohl’s, Wayfair, Overstock and Chewy."

Everyone's in on it, yeah even advertisers and trade associations

The next obvious question you might have is "who are these bad guys?" and why can't we just arrest them and solve ad fraud? Well, two things.

1) There aren't a lot of bot makers. There doesn't need to be. There are vast botnets already in existence; remember those used in DDoS attacks that overwhelm sites with so much traffic that they fail? Fraudsters doing ad fraud just need to "rent time" on these botnets. By buying traffic through resellers, they can specify exactly how many pageviews they want to buy per month (this is like the $1 they invest) so they can sell ads through programmatic exchanges (this is like the $10 CPMs they get for the ads).

2) Everyone else is "in on it." You can now understand why ad exchanges don't want ad fraud to end. Even though they are not necessarily the botnet makers or the fraudsters themselves, their ad exchange makes more money when larger volumes of ads flow through their system. Even if they are aware of the fraud, if no customers are complaining, why cut off a large chunk of your own revenue just to do the right thing? Just look the other way and let it happen. This has proven to be the case over the years as large exchanges, including public ones, are observed looking the other way, despite clear-cut, large volume fraudsters being part of their network.

When I said "everyone's in on it" I meant everyone. Media agencies love to sell "shiny objects" to their gullible clients -- ooh, programmatic media buying, microtargeting, audience segments, etc. Let's buy more. Fraud detection tech companies are not in this to solve fraud. If they did, they'd be out of business. They love fraud to continue; they just report low amounts like 1% IVT so customers think they are doing something and keep buying their services. (Hint: IVT detection tech sucks and can't catch the majority of fraud; I have receipts). Procurement departments at big advertisers love that ad fraud exists (even though they may not know it). Ad fraud delivers the low prices they keep asking for. In some cases, by insisting on lower and lower prices, procurement departments inadvertently helped perpetuate fraud -- agencies had to buy crap inventory from exchanges to hit the low CPMs that were demanded of them. Venture capitalists and transaction brokers rely on ad fraud too; how else would their portfolio companies meet the hockey stick revenue projections they promised? Trade associations like the ANA are not helping to reduce ad fraud; in fact, by putting out ignorant and false narratives, they are perpetuating the fraud too -- "our programs, like TAG, are 'winning' and criminals are being driven underground" (Source: Mediapost -- ANA To Ad Fraud: Drop Dead, We're Winning ). Finally, some advertisers are part of the problem too. The largest advertisers have far too much money to spend in digital. In their effort to "spend it all" they are demanding unrealistic quantities of ad impressions. Fraudsters happily oblige, by using botnets to create enough fake ad inventory to absorb all those dollars.

"Remember the story of when P&G turned off $200 million in digital ad spending? Nothing happened. That should tell you something."

We haven't solved ad fraud because we don't want to. Ad fraud is not a tech problem; it's an incentives problem. Every party up and down the supply chain has an incentive to keep ad fraud going because they are profiting from it directly or indirectly. Put another way, all parties have an incentive to NOT solve or reduce fraud. But you can see it if you looked. I've been looking for the last 10 years, and the latest example comes from just a few weeks ago. Large scale monitoring across publishers shows a categorical drop in volume after January 1, 2022. Did a whole bunch of humans get together and coordinate their behavior across hundreds of thousands of sites and decide to visit at exactly half the rate they did in December? Of course not. Botnet traffic was turned off across those vast swaths of sites to save money since there is perennially lower demand in January.

Everyone's bathed in money from these trickle down economics

Let me wrap up with this trickle-down chart, updated with full-year 2020 numbers for U.S. digital ad spending. Using round numbers the $150 billion spent by advertisers in the U.S. flows through good publishers (10%) and other supply paths (90%). The vast majority of the dollars are spent by the largest brands, and spent through their media agencies. These agencies love to buy digital ads, particularly programmatic ones, because it's high margin for them, not to mention the hidden arbitrage they do on top of that. The money then cascades down through large exchanges, to smaller exchanges, and down to vast numbers of long-tail sites, fake or otherwise. Those long tail sites that can't get enough humans to visit simply buy traffic from traffic sellers/resellers, that are no more than middlemen renting time on vast botnets to deliver as many pageviews as were purchased. Everyone is bathed in money all the way down. Talk about "trickle down economics!"

So What?

The economics of ad fraud are simple. The incentives of ad fraud are also simple to understand. “Show me the incentives and I will show you the outcome.” That’s a Charlie Munger quote about how incentives drive nearly everything. And it apparently holds true in digital advertising too. Experience has shown me, most won't care about the above. But if YOU care, what can you do about it? The answer is equally simple -- buy real ads from real publishers that have real human audiences. Period. (If you need help doing that, please let me know and I will help you.)

A sincere thank you to all my supporters over the years, both public and behind-the-scenes. I've learned from you; I've been challenged by you. Let's go make digital advertising better, together. Now's the time.

Talk with you soon,

Augustine

The "other" 299 articles: https://www.dhirubhai.net/today/author/augustinefou