WE LIVE IN an age of manipulation. An extensive network of commercial surveillance tracks our every move and a fair number of our thoughts. That data is fed into sophisticated artificial intelligence and used by advertisers to hit us with just the right sales pitch, at just the right time, to get us to buy a toothbrush or sign up for a meal kit or donate to a campaign. The technique is called behavioral advertising, and it raises the frightening prospect that we’ve been made the subjects of a highly personalized form of mind control.
Or maybe that fear is precisely backwards. The real trouble with digital advertising, argues former Google employee Tim Hwang—and the more immediate danger to our way of life—is that it doesn’t work.
Hwang’s new book, Subprime Attention Crisis, lays out the case that the new ad business is built on a fiction. Microtargeting is far less accurate, and far less persuasive, than it’s made out to be, he says, and yet it remains the foundation of the modern internet: the source of wealth for some of the world’s biggest, most important companies, and the mechanism by which almost every “free” website or app makes money. If that shaky foundation ever were to crumble, there’s no telling how much of the wider economy would go down with it.
Hwang draws an extended analogy between the pre-2007 housing bubble and today’s market for digital advertising. In the years leading up to the Great Recession, American lenders went wild, issuing mortgages to people who (in retrospect) were unlikely to pay them off. Those loans—the infamous “subprime” mortgages—were then packaged into complex financial instruments that hid the shakiness of the underlying assets. Investment banks and other financial institutions bought into those securities without quite knowing what was in them. When the housing market sagged, it triggered a panic that tanked the global economy.
Just as housing played an outsized role in pre-crash financial markets, so does advertising in the digital economy. Google earns more than 80 percent of its revenue from advertising; Facebook, around 99 percent. Advertising also makes up a fast-growing share of Amazon’s revenue. The global market for digital advertising was $325 billion last year and is projected to grow to $525 billion by 2024. All that wealth is used to fund myriad other ventures—including cutting-edge research into AI and clean energy—that might wither away if the advertising spigot were turned off.
If the financial market of the aughts was dangerously opaque, so, too, is modern internet advertising. In the early days of online ads, a brand would strike a deal with a website owner to host a paid banner. The onscreen space for that image, known as the ad inventory, would be sold by the publisher directly. (The magazine you’re reading right now made the first such transaction, back in 1994.) Today, the process has grown far more complicated, and humans are barely involved. “As they do in modern-day capital markets, machines dominate the modern-day ecosystem of advertising on the web,” Hwang writes. Now, whenever you load a website, scroll on social media, or hit Enter on a Google search, hundreds or thousands of companies compete in a cascade of auctions to show you their ad. The process, known as “programmatic” advertising, occurs in milliseconds, tens of billions of times each day. Only automated software can manage it.
Similar conditions were in place when mortgage-backed securities flooded the market in the early 2000s. These financial instruments traded at prices far above their true value, because the average trader had no idea they were backed by toxic assets. Once the truth came out, the bubble burst.
Hwang thinks online ads are heading in the same direction, since no one really grasps their worthlessness. There are piles of research papers in support of this idea, showing that companies’ returns on investment in digital marketing are generally anemic and often negative. One recent study found that ad tech middlemen take as much as a 50 percent cut of all online ad spending. Brands pay that premium for the promise of automated microtargeting, but a study by Nico Neumann, Catherine E. Tucker, and Timothy Whitfield found that the accuracy of that targeting is often extremely poor. In one experiment, they used six different advertising platforms in an effort to reach Australian men between the ages of 25 and 44. Their targeting performed slightly worse than random guessing. Such research indicates that, despite the extent of surveillance tech, a lot of the data that fuels ad targeting is garbage.
Even when targeting works as promised, and the ads are served to their intended audience, many are simply never seen, because they load somewhere out of sight, like the bottom of a webpage. The rise of ad blocking makes the problem even more acute. Hwang cites a 2015 Adobe estimate that ad blockers deprived online publishers of $21.8 billion in annual revenue, more than Facebook’s entire take for that year. Then there’s the astonishing level of digital ad fraud, including “click farms” that serve no purpose other than for bots or paid humans to constantly refresh and click ads, and “domain spoofing,” in which a bottom-dweller site participates in ad auctions while disguised as a more prestigious one. Hwang cites a 2017 study finding that, between lousy ad placement and outright fraud, “as much as 56 percent of all display ad dollars were lost to fraudulent or unviewable inventory in 2016.”
Despite the extent of surveillance tech, a lot of the data that fuels ad targeting is garbage.
It’s fair to wonder why, if programmatic advertising is such a bum deal, so many brands continue to pour money into it. The reasons are manifold and overlapping. To begin, most of the people responsible for ad spending have no idea where their ads are actually running, let alone how they’re performing, and certainly have not brushed up on the latest research papers. That’s especially true for the small and medium-size businesses that make up the bulk of Google and Facebook advertising customers. I spoke recently with the owner of a successful online audio equipment store who had recently learned, thanks to a chance encounter with an expert, that 90 percent of his programmatic ad budget was being wasted on fraudulent clicks. Most other merchants simply never find out what happens after they send an ad out into the world.
Hwang identifies other structural factors that keep the bubble inflated. The online ad market is laden with perverse incentives to hide the true value of the assets for sale. Advertising agencies engage in arbitrage, buying ad inventory at a discount from publishers and selling it at a markup to their own clients. So, too, do the digital platforms that serve as middlemen between buyers and sellers. There is no independent arbiter. The nearest thing, the Media Ratings Council, includes both Facebook and Google as members, along with other ad tech companies. The council is supposed to set objective standards to measure ad impressions, but in practice its role may more closely resemble that of the credit-rating agencies that slapped AAA ratings on junk mortgage securities.
Hwang moves through this evidence rather briskly, but most people who study programmatic advertising have ended up with roughly the same degree of cynicism. Sinan Aral, a tech entrepreneur and academic who directs the MIT Initiative on the Digital Economy, surveys the research exhaustively and cautiously in his book The Hype Machine. He notes that while “some digital and social media messaging is quite effective,” it’s common for platforms and media agencies to triple (at least) its apparent value by wrongly crediting digital ads for purchases that consumers would have made anyway. Aral calls this “the most widely used shell game in business today.”
THESE PROBLEMS AREN’T entirely new, of course. Hwang cites an adage attributed to the 19th-century businessman John Wanamaker: “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.” But Wanamaker was grappling only with the problem of attribution—figuring out whether the money he spent on a newspaper ad, say, drove sales that otherwise wouldn’t have happened. Today’s programmatic advertising has that issue in spades, plus the extensive problems of placement and fraud. At least Wanamaker could check that his ads had actually appeared in the newspaper.
Had the biggest ad agencies of the analog age—companies like Ogilvy or WPP—gone belly-up in the 1980s, the fallout would have been limited to Madison Avenue. Now the central players are Facebook and Google, with Amazon racing to join them. Those three companies alone account for roughly 10 percent of the US stock market’s total value. Their destiny is bound up with that of the global economy.
What would it look like if the ad bubble burst? Hwang compares today’s units of online ad inventory with the toxic securities of 2007: Both derive their value from an overvalued, hidden asset. (For one, it’s a shaky home mortgage; for the other, a user’s putative attention.) It would be more apt, however, to compare those pre-recession financial instruments with the stocks of companies that make their money from digital advertising. The sale of a moment of an internet user’s attention is a one-time transaction. A financial bubble, however, requires an investment made at time A to prove worthless at time B. A mortgage-backed security is, well, a security: an investment backed by the future value of the underlying mortgages. A share of Facebook or Google stock is an investment backed by the company’s future earnings from digital advertising.
Is that really what we’re going for—a better functioning, more effective market for behavioral targeting?
So if Hwang is right that digital advertising is a bubble, then the pop would have to come from advertisers abandoning the platforms en masse, leading to a loss of investor confidence and a panicked stock sell-off. After months of watching Google and Facebook stock prices soar, even amid a pandemic-induced economic downturn and a high-profile Facebook advertiser boycott, it’s hard to imagine such a thing. But then, that’s probably what they said about tulips.
This is not something to be cheered. However much targeted advertising may have skewed the internet—prioritizing attention-grabbiness over quality, as Hwang suggests—that doesn’t mean we ought to let the system collapse on its own. We might hope instead for what Hwang calls a “controlled demolition” of the business model, in which it unravels gradually enough for us to manage the consequences.
How might that work? Hwang proposes a publicity campaign by researchers, activists, and whistleblowers that exposes the sickness of the online ad market, followed by regulations to enforce transparency. Digital advertisers would have to make public, standardized statements to help buyers evaluate their wares. The goal would be to narrow the dangerous disconnect between perceived and actual value.
The idea of applying stock-market-type regulations to the digital ad sector is having a bit of a moment. The antitrust scholar and former ad tech executive Dina Srinivasan makes a similar argument in a forthcoming paper, and has gotten the attention of at least one member of the House Antitrust Subcommittee. It’s fairly intuitive: A sprawling marketplace representing hundreds of billions of dollars of wealth probably shouldn’t remain an ungoverned free-for-all; and replacing today’s opaque, monopolistic market with a transparent, regulated one might lead to more innovation in ad targeting and more competitive pricing. But is that really what we’re going for—a better functioning, more effective market for behavioral targeting?
Market correction, implemented on its own, won’t eliminate the pathologies of behaviorally targeted advertising: The pervasive surveillance of where you go, whom you know, how often you pee; the redistribution of billions of dollars in ad revenue away from news organizations and toward social media platforms and ad tech middlemen; the ability to microtarget political messaging to nudge swing state voters to stay home. Only legislation that outlaws the business model, or heavily disincentivizes it, will create room for more benign technologies to arise.
It’s a strange thing, the internet economy. The product that generates all the money doesn’t work very well, and when it does work, people tend to hate it. The question is which problem should be solved.