The internet was supposed to mean a whole new world for the business of advertising. Gobs of data let advertisers become wildly efficient in who they target and how they measure results. Consumers also ostensibly win: If you're in the market for a quality winter coat, the thinking goes, you're not going to be annoyed if you see an ad for one.
In this new world, Facebook is on top. It knows so much about its users that it can deliver ads precisely calibrated for virtually any demographic you can dream of, from suburban grandmothers to millennials living abroad. But lately, Facebook has faltered, exposing cracks in the basic assumptions about the superiority of digital advertising—the business model on which so much of the internet has run for the past 20 years.
Last week, Facebook said it found flaws in the metrics it reported to advertisers—the measurements by which those advertisers judge the success of their ad campaigns on the platform. The company said it overstated the reach of Facebook Pages and Instant Articles, as well as its count of referrals to apps from ads. This admission of miscounting came just a few months after Facebook said it had inflated how much time on average viewers spent watching video ads for two years.
Facebook has promised more transparency. But in media and advertising circles, some critics are starting to ask whether they've been spending their money wisely on Facebook. Were they duped into making costly business decisions based on wrong information? Yes, Facebook is still an enormously powerful platform. Close to a fifth of the world’s population checks the social network every day, and all those eyeballs are incredibly valuable to advertisers and publishers. But the revelation that it exaggerated its own reach points to real flaws in the narrative about what Facebook—and indeed, digital advertising on the whole—can accomplish.
Facebook says it has an incredibly robust system that offers up to 220 different ways to measure how well advertisers' ads do. But Facebook's sterling reputation has left advertisers all the more confounded by how Facebook's miscounts could have happened in the first place. “Facebook is supposed to be the best of the best, the cream of the crop, with advertisers spending good money with them,” says Brian Wieser, a media industry analyst at Pivotal Research. “There’s an expectation that you don’t get this thing wrong.”
And yet Facebook did. It overstated app referrals by 6 percent on average by counting not only posts that directed traffic back to the app maker’s website or app but also clicks to view photos or video that kept a user inside of Facebook. Facebook Pages—the pages brands maintain as their home bases on the site—double-counted repeat visitors, leading to greatly exaggerated estimates of the size of their audiences. (The company said page owners should be prepared to see their "28-day reach" fall by 55 percent.) Facebook also said that, due to a math error, it estimated audiences were spending 7 to 8 percent more time reading fast-loading Instant Articles than they really were.
The company doesn’t say exactly how it came to discover these new discrepancies, pointing to “bugs” in the system. But Wieser says it’s likely that Facebook looked into the issue more deeply after finding the mistake in counting that video metric back in October. “Pick your analogy or theory,” Wieser says. “The cockroach theory, where you see one somewhere, then you see more, and say, geez, this must be something systemic. Or the Comey analogy: the problem was brought up then seemed to be settled, but then it comes up again.” Whatever the case, he says, the mistakes definitely amount to a “sackable offense.” (Facebook says it has established what it's calling a measurement council to respond to advertisers' concerns.)
But this still doesn't get at what Wieser sees as the bigger problem: Facebook "grading its own homework." You can’t get a report on Facebook metrics from anywhere else—only from Facebook. And Facebook defines what those metrics are. In the wake of its most recent debacle, the company did pledge to work with third-party industry reviewers, something Rob Norman, chief digital officer of powerhouse media buyer GroupM, applauds. But these reviewers can only retroactively verify Facebook's metrics. They can't get an independent view into the data themselves.
Norman wants to be able to analyze the activity of advertisers and their competitors on Facebook as it already can on most other media, including TV and print. On Facebook, advertisers can see whether a video has been watched and how long it's been viewed for. But they still can't get any information on, say, how often a brand’s rival runs video ads on Facebook. In Norman's view, the information most useful to ad buyers is still not the information Facebook is choosing to release. "I genuinely don’t know whether, if we had the information we wanted, we would be spending more, the same, or less money on Facebook," Norman says. "But what I do know is that I would be able to look at Facebook advertising in the broader context of all advertising."
The data digital platforms provide to advertisers is the medium's key selling point. To be fair, Norman says he never thought Facebook was being intentionally misleading by misreporting its metrics. But when Facebook as a gatekeeper doesn't seem to be offering the right data in the first place, the pitch for digital advertising becomes less convincing.
Still, it’s impossible to imagine running a digital ad campaign today without running some part of it on Facebook. Trying to get elected? You can target ads for specific congressional districts. Selling mattresses? Facebook knows which users just moved. You can target people for their tastes, too: restaurant ads for Italian food connoisseurs, or furniture ads for fans of mid-century modern. Because your history, preferences, location, and network of friends are all part of Facebook’s data set, how could ad targeting not work? The numbers make the company look infallible: in 2015, Facebook pulled in $17 billion in ad revenue, a 49 percent increase compared to one year ago. It boasts 1.8 billion monthly users—more than the population of China.
Targeting is at the core of what Facebook advertising offers. And the narrower the targeting gets, the more expensive Facebook charges for an ad.
But what if the problem isn't the technology but the way Facebook conceives of consumers themselves?
"Facebook somehow contemplates a world where, if I owned a skateboard company, I only show a skateboard ad to men aged 18 to 25," says Benjamin Edelman, a Harvard Business School associate professor who studies online advertising. "So all the other customers—the grandma who wants to buy a skateboard for her grandson, the 40-year-old who wants to relive his childhood—I’d totally miss them. What’s the point?"
It’s "a shock" that so many advertisers buy into Facebook when it only seems to play to people’s worst preconceptions of who customers are and who’s interested in what, Edelman says. Those preconceptions might lead to unsophisticated and inflexible targeting that fails to address other markets advertisers hope to reach. They might even, as a recent ProPublica investigation found, promote racial bias. Targeting does make sense for some---say, a local business owner who wants to let her neighbors know she just opened a new shop in town. But what about the multinational consumer goods companies of the world? You’ll buy Johnson and Johnson’s household products if the store across the street from where you live carries them, Edelman argues---whether or not J&J decided to take out ads on Facebook.
Facebook may or may not be the best place for advertisers to spend their dollars. Either way, many in the ad industry say it's time to retire the idea that digital advertising is closer to perfect than other kinds of advertising. "The idea that digital was this perfect nirvana was a view that was never reflected in reality," says Wieser. If some advertisers were surprised that Facebook's metrics weren't perfect, he says, they weren't paying close enough attention. After all, the very idea of what counts as success in digital advertising is still in flux. Just look at how different tech companies disagree on what to call a “video view.” Does a partially obstructed video view count? How long does a user need to watch it? Should they get to the end of the clip? And what about ad fraud caused by software like ad blockers?
"Ad-supported digital media covers an almost countless number of variations, many of which bear no resemblance to anything that came before," says Randall Rothenberg, president and CEO of the Interactive Advertising Bureau, an industry trade group.
Despite all that uncertainty, ad buyers are hardly ready to abandon Facebook. No matter how unclear the metrics of success on the platform really are, whether through miscounting or simply lack of a long enough track record, advertisers don't feel they can afford to wait to figure it out before committing their dollars. "The pendulum has swung about as far as it can where you can lose your job for not buying ads on Facebook, but you keep your job for not buying ad space in The New York Times," Edelman says. If it does start to swing back, he says, it may be due to incidents like reporting bad numbers to advertisers—the real customers who keep the social network in business. Facebook may be powerful. But as these cracks show, it's not foolproof.
