When an online ad is delivered to a persons computer based on previous websites

Not long ago, I decided to test how much privacy I have online. I cleared the cookies, the bits of code that Web sites leave on my computer to track what I browse and buy, from my two Internet browsers, Safari and Firefox. Then, with my digital past superficially erased, I set out to create two new identities: Democratic Jeff and Republican Jeff.

Safari became the home of Democratic Jeff. I started by spending time on Barack Obama’s re-election Web site and then visited some travel, car and shopping sites to search for flights to Los Angeles, Volvos and Birkenstocks. On Firefox, as Republican Jeff, I went to Mitt Romney’s site and then searched for Cadillacs, flights to Hawaii and diamond rings.

Having created my new digital identities as heavy-handedly as possible, I returned to my usual Web sites. At first, the ads on my favorite Washington neighborhood blog, the Prince of Petworth, were the same on both browsers. But less than two days later, an ad for Mitt Romney suddenly appeared next to a story I was reading on Firefox about Gore Vidal’s burial. When I opened that page on Safari, the ad in the exact same spot was for Catholic University’s master’s program in human resources management.

How did Republican Jeff and Democratic Jeff end up seeing entirely different ads? The answer is real-time bidding, a technology that’s transforming advertising, politics, news and the way we live online. Advertisers compete in an auction for the opportunity to send ads to individual consumers. Each time a company buys access to me, it can bombard me with an ad that will follow me no matter where I show up on the Web.

To dig deeper into my new identities, I visited the Web site of BlueKai, one of the leading online data aggregators. The company’s software enables its customers to sort consumers into 30,000 market segments like “light spenders” and “safety-net seniors,” and this fine-grained categorization helps make real-time bidding possible. According to BlueKai, Republican Jeff is someone who makes between $60,000 and $74,999 a year, lives in Portland, Me., is interested in luxury cars, celebrities and TV, may have bought a cruise ticket, is an ideal candidate to take out a mortgage and a “midscale thrift spender.” Democratic Jeff is someone who lives in Los Angeles, Long Beach or Santa Ana, runs a large company with more than 5,001 employees and cares about advertising and marketing. Neither of these profiles is accurate. Nevertheless, the pigeonholing of Republican Jeff and Democratic Jeff represents our digital future.

Google and Facebook have each been expanding their use of real-time bidding. In June, Facebook announced that it would introduce a new service called Facebook Exchange, which will enable advertisers to send promotions for Spanish hotels, say, to Facebook users who have searched for trips to Spain.

Should we worry about ads aimed specifically at us everywhere we go on the Web and, increasingly, on our mobile devices too? Yes, and not just because the ads can be invasive and annoying. Real-time bidding also makes the online marketplace less of an even playing field, allowing companies to send loyalty points or discounts — or price increases — to individuals based on their perceived spending power. The travel site Orbitz, after learning that Mac users spend 30 percent more on hotel rooms than P.C. users, has started to send Mac users ads for hotels that are 11 percent more expensive than the ones that P.C. users are seeing, according to a recent Wall Street Journal article.

Of course, many consumer breaks are unfair, and we readily accept that the cost of airline tickets, for example, varies from one passenger to another on the same flight. But our consumer profiles are beginning to define us in all of our online interactions, and a result may be that we get different prices at the mall — or different news articles and campaign ads on our mobile devices — based on a hidden auction system that we’re unable to alter or control.

This is the darker side of the online personalization that otherwise delights us on our iPhones and tablets. At the Mobile World Congress in Barcelona in 2011, Eric Schmidt, now the executive chairman of Google, said that mobile devices would soon “do things we haven’t begun to think of,” like storing details of our preferences and tastes and offering location-based suggestions that anticipate our desires and our questions before we’ve even asked them. Of course, a world where Google tells us what we should be doing next is, for each of us, a world of one inhabitant — a place that we never consciously chose to enter and from which there’s no exit. From the perspective of companies like Google and BlueKai, I am not the consumer — I am the product, waiting to be sold to advertisers. Inevitably, some people will be considered more valuable products than others, and they will live in different virtual worlds as a result.

Beginning in the “Mad Men” era and continuing into the early Internet age, TV networks and magazine publishers sold ad space by persuading advertisers that their audiences included demographic groups likely to buy particular products. In the 1950s, it has been said, 95 percent of the success of advertising agencies came from the creative department, which designed the ads, and only 5 percent from the media department, which paid to place the ads. But the rise of independent media-buying firms has made it possible to identify the individuals most likely to be receptive to ads. This is bad news for magazines and newspapers: once advertisers were able to track and reach specific consumers, they became less interested in where their ads appeared and more interested in who, specifically, was seeing them.

This shift is transforming the economy of online advertising. Google still depends largely on ads tied to search. These are based merely on whatever terms you enter at a particular moment: search for “Hawaiian vacations,” and ads for Hawaiian hotels are likely to pop up on the results page. Such ads, which resemble old-fashioned classifieds, produced a vast majority of Google’s nearly $38 billion in revenue in 2011. On the other hand, only 0.1 percent of all display ads, which are more like magazine ads, with text and images that can appear anywhere, are clicked, according to one estimate.

But as consumers flock to devices like smartphones and tablets, the potential of personalized display ads is making them increasingly popular with advertisers. Since 1994, when Lou Montulli, an employee at Netscape, created the cookie as a way of distinguishing online shoppers, it has been possible to track the activities of individual users on particular Web pages. It wasn’t until the following decade, however, that real-time bidding first used cookies to tag individual Web browsers so that their users could be sent display ads at various Web sites. This makes it possible to build comprehensive profiles of users and then conduct an auction among advertisers to show a display ad to targeted users across tens of thousands of Web sites. Google hopes its revenue from ads that are not tied to search queries will grow significantly. In 2010, the worldwide business in display ads was about $25 billion, Neal Mohan, Google’s vice president of display advertising, told me. But, he added, “there’s no reason that $25 billion couldn’t be $100 billion in a few short years” — as the industry delivers more and more ads to mobile phones, tablets, smartphones and even interactive televisions.

Image

When I visited the offices of BlueKai last year, I met Omar Tawakol, who helped found the company in 2008. After studying mechanical engineering at M.I.T., Tawakol went to grad school for computer science at Stanford, where he became obsessed with the idea that data about individual Internet users could be valuable in itself, regardless of where it was collected. “Right now, data looks like black, gooey material,” Tawakol told me at his office in Cupertino, Calif. “Oil was to the industrial revolution as data is to our information economy.” His sense of the potential scope of the marketplace he would help create is reflected in the name he chose for his company: “kai” means “ocean” in Hawaiian.

BlueKai’s customers — which have included travel sites, like Kayak and Expedia, that want to advertise to individual consumers — now track more than 80 percent of the U.S. online population and have created more than 200 million individual profiles based on what we browse and buy online. (By some estimates, there are more than two profiles for every person in the United States.) At the time of my visit, Tawakol told me that over the previous 30 days, BlueKai’s cookies indicated that 38 million people had been to travel sites, and 635,000 of them had plugged in “Hawaii” as a destination. Next, he explained how the BlueKai data exchange then worked. Let’s say you’re planning a trip to Hawaii. You visit a travel site that works with a data intermediary like BlueKai. With the travel site’s cooperation, BlueKai puts a cookie on your computer that records the fact that you have looked up flights from San Francisco to Maui with a seven-day advance purchase. BlueKai’s extensive partnership network enables it to follow more than 160 million people every month who are looking to buy things like cars, financial services, retail and consumer goods or travel accommodations. By sorting users into categories based on our interests and purchasing power — “midscale thrift spenders,” for example, or “safety-net seniors” — BlueKai’s software helps advertisers determine how much each of us is worth following, and at what price. Advertisers for Hawaiian hotels, restaurants, car-rental companies, souvenir shops and so on then place bids starting at one or two cents for each anonymous consumer.

The winning bidder — the Maui Hyatt Regency, for instance — next goes to an advertising exchange, like Google’s DoubleClick Ad Exchange, which conducts a separate auction to determine what the Hyatt has to pay to send you an ad whenever you show up on a Web page that has a relationship with DoubleClick. The Hyatt bids against the entire pool of other would-be advertisers who may not know that you want to vacation in Hawaii and therefore bid less to send you an ad. The automated auction is conducted in real time, which means that as soon as the Maui Hyatt wins the auction, its ad shows up within milliseconds of your loading a given Web page. “What real-time bidding did is to open up a world where the advertising buyer can come in and very specifically, person per person, decide who they want at what price,” Tawakol said. On one occasion after I searched for flights to Maui on Orbitz, the display ad that suddenly showed up at the top of The New Republic home page was for package deals to Kahakuloa Bay.

Two years ago, Tawakol concluded that businesses would rather create their own user profiles and restructured his data exchange accordingly. BlueKai no longer collects and sorts data about consumers; it provides the software that enables Web sites to track their users and put them into market segments. A travel site where you look into flights to Honolulu, for example, will take your profile to a real-time bidding exchange and bid on the right to flood you with ads about Hawaiian hotels, no matter where you show up on the Web. “This is shockingly big, now that people are using their own data,” Tawakol told me. “We’re talking about 80 billion times a day” that ad transactions are taking place. In addition to changing the structure of the data exchange, Tawakol also expanded a separate BlueKai service — its “data management platform” — that allows advertisers to send customers ads or discounts across a range of digital platforms — from the Web to mobile devices and eventually to Web TV.

As Tawakol contemplates the future, he imagines a personalized world that extends beyond users’ Web experiences. “There will be a concept of a unique and anonymous consumer across platforms — online, offline, mobile and digital TV,” he said. “If you’re looking for a trip to Hawaii online, you will see an offer on your mobile phone and your home TV.” Tawakol added that “once we figure out the privacy rules” the tracking will all be connected: “Ads you see on TV will be informed by where you logged in, what you saw online and what products you’re using.”

Joseph Turow calls this ubiquitous tracking and personalization “the long click.” A professor at the Annenberg School for Communication at the University of Pennsylvania, Turow characterizes the process in his book “The Daily You: How the New Advertising Industry Is Defining Your Identity and Your Worth.” A consumer gets a particular reputation and is then inundated with offers online and off, and the results are tracked as he passes through an advertising gantlet — from customized articles (whose writers’ pay is based on how many products their stories sell) to discount offers transmitted to TVs and mobile phones. Some marketers are even looking forward to digital billboards that will flash personalized discounts or pricing the moment you enter a store, after reading your customer profile from your mobile device.

The long click raises obvious privacy concerns. But the privacy threats go beyond troublesome ads. BlueKai’s business model stands or falls on the idea that our digital profiles are anonymous at the time they’re auctioned off. In fact, computers can link our digital profiles with our real identities so precisely that it will soon be hard to claim that the profiles are anonymous in any meaningful sense.

Paul Ohm, a law professor at the University of Colorado at Boulder, argues that this goes far beyond creepiness. Companies can combine hundreds or thousands of facts about us into what Ohm calls “a database of ruin.” For example, by knowing discrete and apparently unconnected facts about you — your shirt color, gait, driving habits and the e-mail font you use — companies could, using algorithms that sort the profiles of hundreds of thousands of people like you, accurately predict what kind of porn you surf. For each of us, Ohm argues, there’s at least one closely guarded secret that could lead to devastating harm if revealed — “a medical condition, family history or personal preference,” he says — and the database of ruin will make that secret increasingly hard to conceal.

Tawakol stresses that the BlueKai profiles contain only generic information. When we were in his office last year, he pulled up his own consumer profile. “It shows I have Comcast, says something about me being in this area geographically and thinks I’m interested in certain books and magazines,” he said.

When I checked the broad outlines of my profile — anyone can do this from home — I found that a single search for a flight to Hawaii on Kayak.com had landed me in the following BlueKai categories: “Economic Spectrum > Midscale: Light Spenders,” “50K-100K Income” and “Age: Retired (65+).” I was also put in the “Safety-Net Seniors” and “Small-Town Retired Couples” segments. As in the case of Republican Jeff and Democratic Jeff, none of these classifications was accurate, although Tawakol noted that they would become more precise the more I browsed the Web while allowing BlueKai cookies on my computer.

BlueKai has emphasized its commitment to privacy by allowing users to block its cookies and to see their profiles at BlueKai’s Web site. Tawakol has also developed a way of guarding mobile privacy by hiding the unique identification numbers on our smartphones. And BlueKai says that its advertising partners can’t identify by name the consumers they’re tracking, and they generally don’t want to. “We stop allowing any trackability after 90 days for privacy reasons because we don’t want someone’s profile to get so full of stuff that it feels bad,” or creepy, Tawakol said. “And not only is the profile anonymous, it can’t have anything sensitive — sensitive financial data, nothing to do with health care, sex, adult material or religious preference.” The Obama administration has not been satisfied with self-regulation, however. Earlier this year, the Federal Trade Commission issued a privacy report that renewed its call on the advertising industry to develop a “Do Not Track” system that would let individuals opt out of Internet tracking more easily than they can now.

The industry initially resisted these proposals, insisting that any restrictions on tracking technology would cripple the economic viability of the Internet. In February, however, a coalition of Internet giants, including Google, agreed to support the idea of embedding a Do Not Track button in Web browsers that will allow users to ask Web sites not to track them. Microsoft has gone even further, angering its competitors and the advertising industry by making Do Not Track the default setting in Internet Explorer 10.

Given the possibility of turning cookies off, should we really be concerned about what some call “follow-me advertising”? In practice, fewer than 15 percent of Internet users turn off or limit their cookies, according to surveys, indicating that most people don’t know they’re being tracked this way or don’t care. But the greater concern is that cookie technology will soon be obsolete as follow-me advertising moves from the Web to mobile devices. Already, companies like Belly offer customer-loyalty programs that link discounts to your smartphone. Using G.P.S. tracking, an app with a “near me” button alerts users to businesses in their vicinity that accept the Belly card. After customers scan the card in a store, Belly sends loyalty points to their smartphones.

Image

Advertising has always been an arms race, with ads chasing people and people finding ways — TiVo, for example — to avoid ads. But once ads move onto our smartphones and, possibly, onto wearable devices like Google Glasses that aspire to project ads directly onto our retinas, avoiding the ads that are following us everywhere will become much harder than simply deleting your cookies: smartphones have persistent identification numbers that can’t easily be reset.

Last year, there was an outcry when Apple and Google admitted that they were storing geopositioning data about users’ locations on their iPhones and Android smartphones. “It’s weird that the offline world is breaking the anonymity barrier at a deeper level, and somehow that seems to be O.K.,” Tawakol said, referring to our growing comfort with mobile devices’ knowing where we are. If smartphones that track location can tell when you’re buying shirts at Sears — and enable you to receive loyalty points or discount coupons for the shirts at the checkout line — it’s difficult to imagine how mobile advertising can remain truly anonymous.

Being put into BlueKai-designated market segments — like “safety-net seniors” or “midscale thrift spenders” — isn’t troubling only from a privacy perspective; it can also lead to tangible economic harms like profiling, redlining and price discrimination. As Turow puts it in “The Daily You”: This “strategy of social discrimination will increasingly define how we as individuals relate to society — not only how much we pay but what we see, when and how.”

Tawakol and other proponents of personalized advertising talk about its empowering qualities, its ability to deliver only relevant ads and offers that will surprise and delight consumers, not annoy them. But profiles that define us forever can also be technologies of classification and exclusion. As Google’s 2011 report on real-time bidding emphasizes, the page views of consumers with “the most value to the advertiser” receive very high bids, while those with “the least value” get no bids at all. Unlike a marketplace where individuals haggle with sellers on equal terms, the new world of price discrimination is one where it’s hard to escape your consumer profile, and you won’t even know if companies are offering discounts to higher-status customers in the first place. Those who worry about privacy may struggle to establish different identities on different devices — creating home and work tablets, iPhones and laptops — but most people won’t make the effort. “Some individuals will be given a digital scarlet letter,” says Jeffrey Chester, executive director of the Center for Digital Democracy, describing an invisible marker that identifies someone as not worth dealing with.

If this kind of advertising actually works, will the spread of real-time bidding and follow-me advertising continue? Some skeptics in the advertising business argue that using real-time bidding to send ads to ever-narrower market segments isn’t cost-effective because it can double advertising costs without doubling revenue. “If you were a toothpaste brand, and you knew your ideal consumer was a woman, 25 to 49, with two kids, you might send ads only to those people and severely limit the scope and scale of the rest of your digital ad plan because you’re overtargeting,” says Tom Hespos, the founder of Underscore Marketing. Hespos, a leading skeptic of data exchanges, recommends that his clients — including large pharmaceutical companies like Pfizer — build their brands with a high-profile ad on Yahoo or by sponsoring a radio station on Pandora, rather than by putting all their energies into microtargeting. Similarly, in keynote addresses at the Interactive Advertising Bureau’s annual meetings in 2008 and 2009, Wenda Harris Millard, then co-chief executive of Martha Stewart Living Omnimedia, compared the practice of selling online ads through data exchanges at the lowest available price with selling “pork bellies” and “shmattes.” Instead of focusing solely on technology, she called for an emphasis on creativity and innovative brand development. But 2009 was the beginning of the real-time bidding era. In light of changes in the marketplace, Millard told me, her views on real-time bidding have evolved. “Over time, things got a little better,” she says. “The truth of the matter is that it is really working.”

Real-time bidding is not only reshaping the advertising industry — it’s also changing the way we consume culture. By 2017, it’s expected to represent 34 percent of all money spent on display ads, up from 12 percent this year. And the process is beginning to change the way we receive news too: as it becomes clear that you click on certain ads when you read certain kinds of articles, some sites will give viewers different news based on the value of their predicted responses. Yahoo has beta-tested a Media Interest Manager that recommends “news for you” based on its predictions of the stories Yahoo thinks you’ll be most interested in reading.

Already, if you’re online and reading The Denver Post, which is a partner of the Daily Me, a company that personalizes news content, and if you’ve been identified as a basketball fan, you may receive articles about basketball as well as basketball-themed ads. In the future, we’re likely to see a rise in companies like the Daily Me and Demand Media, whose stated mission is to “fulfill the world’s demand for commercially valuable content” — paying freelance journalists, for example, to write articles about the health benefits of tropical vacations that can be paired with ads for Hawaiian beach getaways and then sent to users interested in booking trips to Hawaii. The stories we see on the Web, on TV and on our mobile devices could be pegged to the market segments in which advertisers have placed us.

As our experiences become customized, there is more at stake than just discount coupons and deals. There’s also the future of our common culture. As personalization shapes not only the ads we see and the news we read but also the potential dates we encounter and the Google search results we receive, the possibility of not only shared values but also a shared reality becomes more and more elusive. In his book “The Filter Bubble,” Eli Pariser describes the social consequences of a personalized culture, which is the core strategy for Google, Facebook, Yahoo and YouTube — which hope to present us with information that’s so directly relevant to our lives that they can sell more ads to which we’re likely to respond.

As Pariser puts it, “Personalization can lead you down a road to a kind of informational determinism in which what you’ve clicked on in the past determines what you see next — a Web history you’re doomed to repeat. You can get stuck in a static, ever-narrowing version of yourself — an endless you-loop.”

Tawakol responds that it’s possible to create “algorithmic serendipity” that can expose people to products they didn’t yet realize they wanted, creating apparent randomness by sending them ads that their profiles wouldn’t ordinarily receive.

“You might find that people who have a luxury car tend to have a high propensity to buy some kind of biking gear, so a person who expresses a high preference for luxury cars might be a good target for biking gear, even though they don’t yet bike.” But this leaves no possibility for individuality, eccentricity or the possibility of developing tastes and preferences that differ from those of people you superficially resemble.

A world of noncustomized ads — broadcast television in the 1960s, for example — sometimes confronted consumers with ads for products that held no appeal for them. But it was a place where people could define themselves and be surprised by experiences in the future that didn’t reflect preferences they expressed in the past. By contrast, a world of customized ads, news and politics is one where advertisers, publishers and politicians rank and differentiate us. They evaluate us not as citizens but as consumers, putting us in different — and often secret — categories, based on the amount of money they predict that we’ll spend or the votes they predict we’ll cast.

As personalization becomes ubiquitous, the segmented profiles that advertisers, publishers and even presidential candidates use to define us may become more pervasive and significant than the identities we use to define ourselves.