But then, confoundingly, catching virtually everyone quite unaware, and by any logic appearing to be an anomalous dip, Web ad prices started to fall—and then kept falling. Well, not everyone was unaware: companies paying for the advertising were not unsurprised at all.
Now, print advertising has traditionally been priced on the basis of the number of people who see the ad, that is, at a cost per thousand (mille)—CPM. The harder to reach and more sought after an audience, the more you could charge to reach it. The bigger the audience, the less you might charge, but, given mass market size, the more money you might make. The span in print might be as much as a near-$40 CPM for Time magazine, with its 1995 circulation of 4 million, and a $75 CPM for the more specialized Fortune, with a 1995 circulation of 750,000.
And, in fact, at a singular moment in time, when digital readers and analog readers seemed roughly equivalent, CPMs were equivalent too, leading the media industry to believe that the future held equal revenue with vastly diminished production and distribution costs (for magazines and newspapers, no printing and delivery).
But then, relatively quickly, you could begin to see the difference between traditional media with its fairly specific attention demands—television showed one show and one ad at a time, magazines and newspapers a page at a time—and a Web site, a new nonlinear world, a hodgepodge, a mess, an assault.
Digital advertising promised, at last, a way to measure the effectiveness of ads, especially if the ads directed readers to the brand’s Web site. You could finally advertise your product to a million people and see how many of those million bought it. The answer was: disturbingly few.
Hence, the fall began, a fall vastly greater than just an adjustment for the lower cost basis. It was a fall of cataclysmic size—from average magazine highs of $30 to $50 per thousand to an average close to 10 percent of that (hence, in the new parlance, analog dollars to digital dimes) to an undifferentiated long-tail CPM of pennies. The fall implied, at least to anyone not in the strictest denial, that there was a qualitatively clear difference in behavior and in attention and in response between the digital audience and the traditional audience.
What’s more, there was so much of this new audience and its accompanying ad space.
It was a pivotal economic breakthrough moment from the buyer’s point of view. In the past, the price of media reflected a scarcity premium. The price was bid up in sought-after properties (that is, properties that could deliver a sought-after audience) because there was limited supply. Television had a fixed number of spots. Magazines and newspapers, even thick ones, had a maximum number of pages (although print, compared with television, could more easily expand, hence one reason it was generally cheaper to buy). But Web sites had unlimited space.
What’s more, in the Web world, advertisers were not even limited to a given Web site audience. Rather, suddenly there was a growth business in middlemen audience aggregators called, ever eager to be associated with television, “networks.” Google offered the biggest such network (DoubleClick was arguably the first Web-wide ad server—and purchased by Google), but there were myriad others, specialized to greater or lesser degrees, each with “inventory” available through the almost unlimited number of Web sites looking to make additional or marginal income from advertising sales.
It was a curious devil’s bargain. In the past, the barrier to entry in the media business was that you not only had to have an audience but you had to have the wherewithal (a sales force) and the credibility (a brand) to sell that audience. Now, you didn’t. Any network would take any amount of inventory you could give it—albeit for pennies on the traditional media dollar. Still, this revenue opportunity, as discounted as it might be, also fueled the ever-marching increase in inventory, further lowering prices.
This was then compounded by an additional form of ad sales called programmatic buying. The promise, and to a certain degree the reality, was that the audience that you paid The New York Times a premium to reach (because there was only one New York Times audience) could be assembled down to a demographic iota outside The New York Times—and at a dramatic discount. Hence, the already low CPMs at the Times became even lower outside of the Times, forcing the Times itself to lower its rates in order to compete with its own demographic, no longer anchored to the paper but free-floating in digital time and space.
Suddenly there were “buying desks” bidding for and selling an ever-expanding inventory.
The Web became a rare form of auction in which, by the math of unlimited supply, prices reliably went down instead of up.
In order to continue to advance, or at least not to shrink, almost every advertising-driven digital business had to turn to some form of traffic pumping.
This included a wide variety of new techniques—search engine optimization and then social network gaming and traffic loop aggregators—creating massive, even hyperbolic, traffic inflation.
The threshold scale in 2010 of approximately 10 million unique visitors per month needed to secure an RFP from a big media buying agency rose to 50 million uniques, or in some cases, 100 million by 2014.
Advertisers have long accused media of servicing distracted audiences (bathroom and fridge breaks in television, inattentive leafing through a magazine), but compared with digital, traditional media was something of a theater with a specific mind-set and focus. On the Web you had an audience that was as though in the middle of a busy street, its attention caught by sudden random movements, loud noises, screeching cars, ugly or comely passersby. The traditional media audience was there by choice more or less; this new audience was, more often than not, dazed and confused. Digital media thrived on distraction (distraction distracted users from focusing on how feeble it largely was) instead of limiting it.
And that is the positive side. The vastly more negative interpretation is that much of this audience wasn’t an audience at all. In one study as much as one third of Web traffic was found to be the product of click fraud, that is, robots were the audience, many emanating from offshore servers, producing the clicks that sites were selling to advertisers (the Association of National Advertisers estimates that a quarter of digital video impressions are fraudulent). Some brands had taken to insisting on an arbitrary discount on the traffic they were buying by 50 percent or more.
And then came mobile—a sea-change shift to mobile—with its smaller screens, greater distractions (users were, literally, in the street), and another quantum drop in advertising prices.
9
EXPLAINING PROGRAMMATIC ADVERTISING
Jay Sears got his first digital job in 1994. Out of college for a few years, he had been working for an old-line PR firm and then wandered into a start-up Internet company borrowing some space on the same floor with a question. He was a junior person on the Pizza Hut account and his question was about promoting pizza online. Twenty years later, Sears was one of the senior executives at the Rubicon Project, which declares itself “on a mission to automate buying and selling for the global online advertising industry.”
In pleated khaki pants and blue blazer, Sears continues to look like a media salesperson from the greatest days of the business. But in a sense he has gone from someone like himself who would have once been representing the effectiveness of media—a highly focused audience on highly limited media space—to representing the ineffectiveness of it: a vastly larger, unimaginably more dispersed audience, with unlimited media space.
Sears, who might have once been the marketing face, the personal relationship at the heart of the media business—and he still maintains a salesman’s bonhomie and good cheer—is now the opposite: he’s a leader in a business proposing to remove the salesman from media sales.
In many ways, of course, it was the salesman, as part of a classic boys’ club, who maintained the price of media. Media space held its high price because there was a limited amount of it and because a high price was good for the media that sold it an
d the ad agencies that received a commission on the sale when they bought it for willing brands.
But digital media turned advertising into an existential issue: there is just too much advertising space. In effect, the audience is too large. Not only is there too much to sell, there is too much for a buyer—for the advertiser paying for this space—to see. In some sense, digital advertising exists more in theory—that is, it must be somewhere—than in practice.
Sears had to face a salesman’s most difficult task: how do you sell what there is too much of?
That is as good a way as any to think about “programmatic” ad sales, the sales methodology that is largely resistant to explanation because it encompasses so many strategies, methods, and intentions, and seems pretty much willing to promise to be anything anyone wants it to be—and that seems certain to be the future basis of all digital media ad sales and to erode ad prices wherever it’s implemented.
That last point is also part of the misunderstanding. While programmatic has certainly lowered the value of digital advertising, programmatic sellers argue that publishers should hire them to bundle their audiences because this will raise their prices.
In a programmatic world, you sell your audience—your clicks, your views, your uniques (or a part thereof)—at a wholesale price to, in effect, a trading desk, with a buy side and a sell side (this is a new unit—a “desk”—at every major advertising and media buying group). The sell side, having bought your traffic, then hands it across the desk to the buy side, which runs a variety of automated trading programs that can deliver preset demographics at preset prices to a client’s ad (both buyer and seller can put in a preset bid and ask).
The innovation here is that, as in a stock exchange, there is a constant auction for an audience—that is, really, for a bundled audience, a set of demographic slices aggregated from Web sites far and wide. These audiences are, in a sense, like mortgages, which in the early years of automated finance first started to get bundled together (and then were sold to a consortium of buyers). The other point is that programmatic sellers advanced the notion that you didn’t need a particular site or content type or publisher or media entity to reach an audience. It’s not the media brand, but an audience’s demographic profile, that you are buying.
Digital sellers have long argued the breakthrough efficiency of this way of reaching an audience, but in fact it is an old media form. In the great days of billboards, when American cities and highways were a lush commercial landscape, outdoor advertising, among the cheapest per-eyeball deals you could ever hope to find, were largely sold in packages—you’d get x number of passersby on x number of more or less random billboards for a preset price (the billboard locations defined your demographics).
Anyway, a demographic, as opposed to a self-selected audience (that is, someone specifically choosing to visit your Web site), increases the supply of eyeballs and hence lowers their price. There is a finite amount of New York Times readers (although, of course, in the digital world, not that finite) but a much greater number of New York Times–type readers (a world of New York Times–type readers who rarely read The New York Times). Obviously if New York Times–type readers are trading at a meaningful discount to actual New York Times readers, that might reasonably affect the price of the real thing. And, because many New York Times readers, most coming to the Times through search engines and other referrals, are not New York Times readers in the old sense anyway, their value might reasonably be iffier than it once was.
In the beginning this kind of free-floating inventory was supposed to just be a minor part of what a site might sell. In publishing terms this was remnant space, the leftover stuff. But the amount of digital leftover expanded because the amount of inventory increased and, what’s more, premium prices—given the advances of programmatic in the market—and remnant prices began to converge. But never fear, because programmatic sellers were then saying that, with new innovations and enhanced technology, they could now use new targeting methods to raise prices that had fallen. And anyway, while the prices were low, programmatic could give you a guaranteed floor (albeit, of course, a low one), which was better than no floor.
But let’s be clear about the essence of the innovation: unbundling the audiences from the brands that had assembled these audiences, and repackaging them as stand-alone unbranded entities, commodified the product, and hence inevitably lowered its price. There were suddenly many entities trying to sell the same commodity, which, with ever-growing digital supply, would force the prices down even more, and, possibly, ever more.
Part of the digital rationale for programmatic is not only that it does provide a greater efficiency but that it would, precisely because it was efficient, come to television too. This was a different argument from the digital position that said digital’s price would go up because it could target more efficiently than nondigital media; rather, now the argument was that digital innovations would similarly force the price of all other media down.
There have been, in several heady years of programmatic growth, confident predictions of how much of the ad market would become programmatic—in the telling, practically speaking, all of it.
This was true and yet not true. The true part defined—including in this definition virtually all of digital—a hopelessly glutted and overexpanded advertising market of largely undifferentiated inventory. Buying space in this market was not art, it was process. Programmatic was a better process, one that offered a greater level of consistent control of marketing message as well as measurement of results across all media platforms, and, too, it was one that ought to work as well for undifferentiated television supply. In fact, vast parts of the television day have large amounts of cheap inventory on no-name shows that reach a random audience, in which, as with digital, it’s a buyer’s market, any price better than no price (after all, once the spot airs, it is forever gone).
But there was another television market that didn’t function like this at all. In fact, television has always been a cleanly and cleverly cleaved market, between the junk and the precious, between a glut of commodity space and scarce premium, with the latter providing most of television’s profits. In essence, that’s what prime time defines.
Even in the most aggressive discussions of programmatic, sought-after television, prime-time television, hit television, a-seller’s-market television, and hence most of the value in television, was artfully excluded from the coming promise of programmatic efficiency.
“Nobody can buy, for example, any over-the-air broadcast inventory in these marketplaces [the programmatic marketplaces]—or, of course, anything that is sold during the up-front where ad inventory on new shows is sold at a discount in order to hedge against potential failure, then sold at a large markup in ‘scatter’ after the season’s hits have been established,” said Sears in an interview, in an acknowledgment of television’s particular economics and the nature of its exclusive product. “It’s just not a system that’s compatible with the rapid-response ad exchange—and it’s how networks amortize the massive costs of programming. It’s pricey to option promising properties, produce pilots from a few of them, send a choice few to series, and then give everyone a raise when one or two shows prove to be hits and take a write-off when others tank.”
In some sense, the advances of programmatic buying have served to separate the new and efficient—with all its message control and accountability—from the old hat and clumsy, television’s old-boy system of backroom deals.
And yet what it also does, in some larger and ironic way, is to further define the dual advertising markets: the downscale market, of commoditized digital audience and junk television, and an upscale, luxury, exclusive television market. The former is bought largely as a pricing function—and with downward price pressure in an ever-expanding market. The latter is a product of limited supply with ever-rising prices.
And that, to a great extent, helps answer that inexplicab
le and frustrating question for digital people as to why television advertising hasn’t followed the American audience to its digital destinations—digital has defined itself as lower-end junk.
10
THE ADVERTISING CURVE
The ultimate end of the media world (or, broader, the modern commercial world) as we know it probably began more recently than with the advent of the Internet. It likely began with the introduction of the DVR in 1999. Both TiVo and ReplayTV (with Marc Andreessen as an investor) launched at the Consumer Electronics Show in Las Vegas in 1999. The premise was to give a viewer control of the TV programming schedule, quite groundbreaking in and of itself, but even more earth-shattering, to provide the technological wherewithal to bypass advertising. Voilà. The media paradigm had been disrupted.
There commenced a long rearguard action to delay the inevitable, but soon enough it was hard not to come to terms with an ultimate nightmare scenario of a world that would no longer willingly tolerate advertising.
Even now, with television’s continuing hold on high-priced advertising, everybody is well aware that it is possible to watch as much television as you’d like without seeing any advertising at all.
Likewise, more and more when you do come upon television advertising, instead of its being a seamless part and parcel of the television experience, a minor toll, it can appear as a quite unnatural and discordant, even fairly incomprehensible, moment—a broken synapse. In other words, there is a growing advertising-free world from which it might not be possible to return to an advertising-ubiquitous world, even for brief visits.
This is partly technology and pricing. Technology enables the unbundling of content from ads (first, rebelliously, through the TV ad-skipping devices, and then officially in OTT—over-the-top content—venues) and new payment options, enabling the consumer to directly pay for content. But it also would be ignoring the obvious not to note that the effectiveness of advertising, that is, traditional, high-margin, display advertising—that builder of brands and creator of desire—has reached some sort of plateau or even a level of net negative impact among jaded media consumers. Every major advertising holding group, those worldwide collections of agencies, marketing services companies, and media buying intermediaries (all with their programmatic buying desks), sees almost all of its growth coming from parts of the world that have only recently become consumer societies. No surprise that advertising tends to work in those markets in the manner it worked here at the dawn of consumer time—it’s a cost-effective way to enhance the desire to buy.
Television Is the New Television Page 6