Who wants to clean up ad tech?

(Note: As always, the below represents my personal views and not those of my employer or the industry at large.)

Within the ad tech industry, the above infographic is legendary. Each year a new version is released, showcasing the explosive proliferation of companies within the digital advertising landscape. Each named grouping ("DSPs," "Creative Optimization," "Ad Servers," etc.) represents a different service provided within the ecosystem, and generally in each of those categories there are numerous companies competing for market share.

The best way to think about this chart is as a money flow, from left to right. On the far left are marketers, or advertisers: the guys who hold all the money, in other words. (This is also known as the demand side.) On the far right are the publishers, or web sites: the people who control the ad "inventory" where advertisers can display their messaging. (This is also known as the supply side.)

Estimates vary widely, but for every dollar spent by an advertiser on the open web (that is, not on closed platforms like Facebook, Instagram, Google Search, Twitter, Pinterest, etc.), anywhere from around twenty cents, to one-third, to over half of that dollar disappears before ever making it to the publisher. Where is all the rest going? Well, to everyone in the middle of that infographic. A single advertising transaction -- that is, one ad impression shown to a single user on the Internet -- can involve a dozen or more companies in mere milliseconds, each of them taking their respective micropayments.

In any marketplace this complex, what you'll start to see is a diffusion of accountability. The more nodes that exist between the two endpoints (advertiser and publisher), the harder it will be for anyone -- the advertiser especially, who controls the budget -- to isolate and hold accountable problematic actors. This is known as the principal-agent problem. (The very fact that no one is even sure how much of each ad dollar gets siphoned off by middlemen underscores the opacity of the marketplace.) Some estimates place the cost of digital ad fraud at $30 billion annually.

Given the sheer enormity of that number, the obvious question then becomes: who cares? As in, who is incentivized to eliminate the fraud? This is a question I've become more and more interested in understanding. But it's perhaps easier to answer the inverse question: who's invested in the continuation of digital ad fraud?

Generally, you'll find your answer on the right-hand side of the above infographic -- that is, in the direction of the money flow. Fraudulent publishers, networks, mobile app SDKs, and other ad tech vendors seek to exploit the complexity of the online ad ecosystem in order to send large amounts of traffic to bogus sites they own or are affiliated with. Buzzfeed's Craig Silverman recently unearthed a network of fake sites designed to collect advertisers' dollars by masquerading as actual local news sites. My employer uncovered a massive Android app network last year that fraudulently played invisible video ads without the end user's knowledge. The FBI and DHS arrested members of a widespread ad fraud scheme in 2018. The list goes on and on. (In one of my personal favorite examples, Bitcoin miners even hijacked the ad supply chain to run distributed mining operations via JavaScript code on unsuspecting users' computers.)

It's obvious why these players would be invested in continuing fraud: they're making money from it. But what about everyone else? What about, for example, supply-side platforms (SSPs) and demand-side platforms (DSPs)? What about advertising agencies?

This is a complicated question to answer, in large part due to the reigning revenue/cost model of online advertising: cost per mille (CPM), meaning the cost per thousand ad impressions delivered. Many of today's ad tech players are paid out on the basis of this metric or a variant of it. Yes, there are many Software as a Service (SaaS) contracts, but even these are generally tiered by volume, meaning that their value derives indirectly from impression volumes as well.

This matters because, in a world where most vendors' revenues are closely tied to impression volume, the possibility that a significant portion of that volume could be fraudulent becomes a much less intriguing mystery to solve. Why devote resources to investigating something that may reduce your revenue? (As Upton Sinclair memorably explained: "It is difficult to get a man to understand something, when his salary depends on his not understanding it.") There is even a potential for this mindset to take hold amongst companies and consultants that specifically market anti-fraud or verification services. On a localized level, it is within their interests to reduce the incidence of fraud within their clients' ad buys. But on an industry-wide scale, an ecosystem free of ad fraud would mean the end of their business (or at least the portion of it dealing with fraud).

As a thought experiment, just imagine how bizarre it would be if the banking or credit card sectors were saturated with companies devoted to finding and eliminating fraud, and imagine further that fraud persisted in a significant percentage of all credit card transactions due to such dizzying layers of complexity on every transaction that no single person in the entire industry could fully grasp it. That is essentially the status quo for today's digital ad economy.

Or think about an extremely simple analogy from the offline world: you pay the neighborhood kid, Jay, $20 to mow your lawn. There's direct accountability there. If he tries to mow the lawn but instead shaves uneven patterns into the grass because he's woefully unfit to be anywhere near a lawnmower and honestly only took the job to help pay the rent over the summer to complement his unpaid congressional internship, you'll fire him. (If that sounded oddly specific, well, I'm sure it's not autobiographical.)

On the other hand, say Jay does an incredible job, not just with your lawn but with all the neighborhood lawns, and thus needs to hire several third-party lawn-mowing contractors to fulfill all the demand. Those contractors, in turn, each have several mowers on staff. Sometimes a mower calls out sick and hires a substitute of his own to cover his shift. The supply chain has now become noticeably longer: you (lawn owner) -> Jay -> third-party contractor -> staff lawn-mower -> substitute lawn-mower.

Now think about the incentive structure. Does the substitute lawn-mower really care if your lawn is well-mowed? Not really. He cares that the regular lawn-mower (the guy who's out sick) pays him for his work, and he's going to exert the minimum amount of effort possible to make sure that happens. How about the regular lawn-mower? Does he care whether you, the owner of this lawn, are happy with the job? Again, not really: as long as his boss, the contractor, is OK with the job he (or, actually, his substitute) did and pays him for it, he's all set too. In general, the farther away in the chain an "agent" is from the "principal" (you), the less they care about the final outcome.

But this still leaves one glaring exception: you, the guy that owns the lawn. Irrespective of everyone and everything else, you care if the lawn is mowed correctly. But when it's not, how do you know where in the chain everything broke down?

This is where today's online advertisers often find themselves. Other market participants -- from publishers to ad networks to SSPs to DSPs to DMPs to measurement companies to trading desks to agencies -- may have convoluted incentive structures based on 1) who pays them, 2) who they pay, and 3) how narrowly success is defined for both sides of those transactions. The advertiser's incentives are theoretically far simpler: more revenue and sales. Right?

Well, mostly. But some of the world's largest advertisers operate as a sort of microcosmic ecosystem of their own, with various C-suite executives and global departments jousting for internal visibility and increased budgets. Even here, inside the belly of the beast, the principal-agency conundrum can rear its ugly head. Furthermore, even if you suspect that a certain advertising strategy is unwise -- say, dispensing millions of dollars in brand advertising to remind people that your company exists -- you may be forced to do it anyway because your competitors are doing the same and you can't afford to lose attention share. (Picture what would happen if Pepsi stopped advertising entirely but Coca-Cola kept going.) "No one ever got fired for choosing IBM," as they say.

In short, for such a large industry, there are surprisingly few checks and balances making sure things aren't running off the rails. Which is one of the reasons you often end up with headlines like "Procter & Gamble's Best Sales in a Decade Come Despite Drop in Ad Spending" and "Chase Had Ads on 400,000 Sites. Then on Just 5,000. Same Results." In such a complicated market, it's hard enough to figure out when something's working that weeding out the bad actors often ends up being relegated to specialized vendors (in a best-case scenario), grifters, or no one at all.