Marketers are increasingly using databases to determine whether to consider particular Americans to be targets or waste. Those considered waste are ignored or shunted to other products the marketers deem more relevant to their tastes or income. Those considered targets are further evaluated in the light of the information that companies store and trade about their demographic profiles, beliefs, and lifestyles. The targets receive different messages and possibly discounts depending on those profiles.
This work of creating reputations from profiles was at the heart of a 2010 PowerPoint presentation that Publicis, a French multinational advertising and communications company, made to some of its clients. Below the title “Why We Are Betting on Ad Exchanges” was a phrase that the agency offered as the basis for its decision to devote its energies to the new approach to advertising spending: “Key benefit: Buy only the impressions that match.” To illustrate how the company would help its clients succeed, a slide displayed several concepts, linked with arrows, leading to the ultimate goal: Publicis would draw on its expertise in “Insight,” “Media,” and “Technology” to achieve “No Waste” advertising for its clients. In other words, the presentation suggested that marketers could finally direct their enormous investments to reach only those customers who mattered to them. Publicis would create target segments based on its clients’ data about best customers. It would then bid in the exchanges for the right to reach people who fit those profiles. They would receive the client's messages, and the audience reached would be free of waste—consumers the client didn't care to reach.
This approach is by no means Publicis's alone. It is becoming standard among media buyers in the top firms as well as many of their smaller competitors. The system described in Chapter 3 is running on all cylinders. Wide-ranging data points indicating the social backgrounds, locations, activities, and social relationships of hundreds of millions of individuals are becoming the fundamental coins of exchange in the online world. Violation of privacy as the law defines it is not at issue here. The companies that deal in Americans’ data know the rules and have learned to hew carefully to the letters of those rules, if not their spirit. Still, the personalized use of the data raises questions about the ways privacy is currently defined in the law and in the advertising industry.
Moreover, circumventing the spirit of privacy has led the advertising business to another social dilemma. The quiet use of databases that combine demographic information about individuals with widely circulated conclusions about what their clicks say about them presents to advertisers personalized views of hundreds of millions of Americans every day without their knowledge. Marketers justify these activities as encouraging relevance. But the unrequested nature of the new media-buying routines and the directions these activities are taking suggest that narrowed options and social discrimination might be better terms to describe what media buyers are actually casting. Despite qualms, publishers are warming to the idea of customizing news and information based on the characteristics of people that advertisers would want to approach. Their aim is to increase the number of clicklike activities on their sites and related engagement measures that will allow them to raise their prices at a time of plummeting CPMs.
Many cookies used for this purpose don't last more than a few months because individuals and browsers delete them. Some marketers that want to have control over the deletion process have been adopting ways beyond the standard cookie to keep persistent knowledge about people. Other marketers believe, though, that most of their cookies last long enough for them to engage target consumers with messages specifically tailored to what they know about them. Regardless of the life of a particular cookie or other anonymous identifier, much of the information about individuals endures and spreads beyond individual marketers, as this chapter will show. That is because a small number of data-collection companies continually and anonymously add basic personal information to the cookies they insert on the computers of hundreds of millions of people. Even with short-lived cookies, then, the media-buying system is laying the technology and philosophy for cultivating enduring characteristics of people across the Web. Replacements for the cookie that substantially extend the ability to identify people take this notion of enduring identities quite a bit further.
Advertisers have been way ahead of publishers when it comes to the interest and ability to personalize material for audiences. Publishers have been lagging partly because implementing personalized material with news and entertainment is more difficult than it is with advertising, and also because executives have worried about negative audience response to content personalization without their knowledge and the ability to control it. Important recent changes in the news-and-entertainment environment, however, have made content personalization based on data collection inevitable. But that story is for the next chapter. Here we will focus on how marketers have enforced the target-and-waste strategy on the Web. Publishers, eager for advertising revenues, are deeply involved in two ways. As we saw in Chapter 3, publishers provide marketers with much of the data used for targeting. And, as we'll see in this chapter, they also help keep the audience in the dark about what is actually taking place.
The social and consumer discrimination that defines personalized advertising results from three converging developments: advertising practitioners’ infatuation with data about online audiences, the rise of companies that can provide that data in a readily accessible form, and the growth of technologies that can selectively serve advertising to individuals based on the data associated with them. Underpinning this massive enterprise is contemporary advertisers’ belief that the best way to engage potential customers with a brand is to follow them with messages that are customized so as to be as relevant as possible to their behaviors, backgrounds, and relationships.
The notion that in the twenty-first century potential customers ought to be pursued with personalized blandishments and appeals represents a new twist on a late twentieth-century understanding of customer relationship management (CRM). In the 1980s and 1990s, CRM proponents preached that it was far more expensive to enlist new customers successfully than to retain and cultivate the best current customers. CRM gurus, notably Don Pepper and Martha Rogers, further argued that new media technologies would allow for mass customization—the efficient personalization of advertising and public-relations messages that would keep customers feeling good about the brand through their understanding of its utility for them.1 The twenty-first-century version of this CRM philosophy holds that databases and other technologies allow for reaching even potential users of the product—individuals who will act like best customers once they are brought into the fold.
That is certainly the view of Acxiom, a firm that catapulted to the top of Advertising Age’s list of the largest marketing-communication agencies in 2009 largely because of its expertise in producing data for the purpose of targeted advertising. Based far from Madison Avenue, in Little Rock, Arkansas, Acxiom touts its ability to help “clients strengthen relationships with existing customers—and initiate the right new customer relationships.”2 The company offers a suite of digital marketing services involving strategy, creative, targeting, and measurement. These services are based on the premise that “expertise in building, maintaining and delivering the most accurate and valuable views of prospects and customers—also known as Customer Data Integration (CDI)—is the single most important ingredient in optimizing every buyer-seller interaction.”3 That activity, Acxiom insists, must take place across digital platforms. “Today,” the company's website proclaims, “brands can gain better control of their advertising and marketing across multiple types of addressable media and concentrate on their ‘best’ prospects and customers through: Display ads, social media, interactive TV and websites.”
Procter and Gamble, one of Acxiom's clients, adheres to this logic.4 Challenged by competitors, concerned about economic downturns, and motivated by a long-standing interest in using digital media to maximize its return on media spending, the world's largest advertiser has in recent years been focusing on more efficient ways to cultivate customers and potential customers. As if following an Acxiom game plan, some of these activities represent an effort to strengthen ties to current customers, and some are attempts to find new ones. “Fan pages” for brands such as Tide on the Facebook social-media site are prime examples of the first category.5 (Facebook encourages what it calls “Public Profiles” for just this purpose.) So are P&G's websites for specific products such as Pampers (PampersVillage.com) and Mr. Clean (MrClean.com). In seeking new customers, P&G sends bloggers free product samples with the understanding that they will write about them if they like them. The company also collaborates with NBC-Universal on websites that target lifestyle segments (Petside.com and DinnerTool.com) and age segments (Lifegoesstrong.com, for baby boomers).6
P&G lurks in the background of the NBC-Universal sites, perhaps suspecting that they will have greater credibility without the company's name featured visibly on it. In fact, it's hard to tell that P&G is connected to them at all. Ads for a first-time site visitor do not highlight P&G products. Moreover, although the Web pages note NBC-Universal's ownership quite prominently, disclosure of P&G's involvement turns up only on the privacy page. There visitors who click beyond the superficial privacy assurances to the more detailed version will learn that Procter and Gamble has an interest in personalizing its digital encounters with them. A preambles states that “we may collect information about you from a variety of sources, including information we collect from you directly; information we collect about you when you visit our sites, use our services, or view our online advertisements; and information we collect about you from other sources (where permitted by law).” Among the reasons it offers for collecting this information, P&G explains, is its desire to “develop and provide advertising tailored to your interests” and “where permitted by law, provide you with customized, unsolicited offers and information about P&G's products and services through postal mail.”7
The amount of knowledge P&G reserves the right to glean about the customers it encounters online is stunning. Three sections lay out what it collects. The first refers to information P&G requests when people register for one of its sites or fan pages. The long list includes name, e-mail address, postal address, phone or cell phone number, age, demographic information, and “other information about you and your family such as gender or product use preferences and/or behaviors.” The second section notes “information we collect when you visit our sites, use our services, or view online advertisements.” Among the listed items are the links the visitor clicks, the visitor's IP address, the site the visitor went to before coming to the P&G site, P&G e-mails that the person opened or forwarded, and P&G offers or links that the person accessed via e-mails. The third section includes information the company “may obtain about you from other sources, including commercially available sources, such as data aggregators and public databases.” Listed there are data that would have been obtained had the person registered on a P&G site (for example, name, address, age, marital status, and number of children) as well as information that people would not likely share or even know themselves—such as purchase behavior, the person's activities both online in blogs, videos, and discussion groups, and off-line, during in-store shopping.
An executive at an agency that markets some of P&G's brands suggested in a 2010 interview for this book that P&G isn't as aggressive at tailoring ads as the privacy policy might imply and that the disclosures were primarily “forward-looking” statements designed to protect the firm when it does want to use the data in that way. That executive doesn't work for Acxiom, however, and P&G's quiet ownership of the sites and its privacy policy seem right out of Acxiom's playbook. In 2008, Advertising Age reported that “Procter & Gamble, long accustomed to a bombard-the-masses-with-heavily-tested-ads strategy, has been working on better personalizing the consumer experience.”8 It seems evident that for P&G a primary purpose of these sites is to create opportunities for inserting cookies in visitors’ computers to learn what they do on that particular site, on other NBC and P&G sites, and across the Web through cookie matching. In this way P&G can serve these visitors messages—ads, coupons, articles—that leverage the company's understanding of them to promote particular products. These exercises in online personalization may indeed pale in the face of P&G's huge expenditures on mass-audience vehicles such as network television. But with these relatively low-cost activities millions of people experience P&G's concept of them in ways that reflect the company's best interest. So, for example, a mother with a young baby who uses Pampers might receive ads reinforcing that use. She might also receive a series of online discount coupons for P&G products to parallel the growth of her child and Procter and Gamble's desire to guide her purchases related to child-rearing.
For the Ford Motor Company, too, targeted personalization is a central marketing strategy. Andy Pratkin, the digital director of WPP's “Team Detroit” media-buying agency, noted in an interview for this book that his fundamental concern when reaching customers is discovering how these prospective customers learn and how their perceptions are formed. Pratkin argued that the digital environment makes it possible to follow an individual across a gauntlet of brand-related mind-sets, from awareness to initial attitude to various attitude shifts. This sequence of activities reflects what marketing students learn to call “the funnel”—the different stages through which consumers move toward purchase (or not). Pratkin noted that his group could now track individuals as they lurch through these steps. He added, “If we can fit into that dynamic”—that is, if through serving personalized ads Ford could make itself a natural part of a person's car-search process—“we can fit into her.” Success—the sale—would develop organically. “It's kind of like dating,” he said.9
According to Pratkin's logic, then, carrying out a successful dating ritual between a person and a brand requires good matchmaking. That is what Publicis's announced commitment to “buying only impressions that match” means in practice. An extension of this logic points to marketers’ creation of silos of messages surrounding individuals they want to target over time. Consider a few of the strategies that Team Detroit used to cultivate online audiences for the Ford Motor Company's advertisements during 2010:
• It purchased ads on many automotive sites that planted Ford cookies in the computers of visitors to the pages where the ads appeared. This “tagging” of computers by cookies allowed Ford to serve individuals different messages and photos depending on where their clicks positioned them in the car-purchase funnel. “We do a fair amount of tagging,” said Pratkin during the interview. “If people are exposed to our ad, we study their behavior, and if they come to our site, then we retarget them and try to entice them back… . If you go to [the] Kelly Blue Book [website], we serve an ad. We'll know if you didn't do anything. Next time, we'll serve you a different ad. You go to our site, and if you do something and you go off our site [to one of Ford's advertising networks], we serve you other ads.”10 The technology often assembled the advertising automatically, on the fly, to match what the cookie data said about the person's car-searching routines and the previous commercial messages that had been served to the individual.
• In other internet buys, Team Detroit tailored its ads on the basis of people's ages, incomes, geographic locations, and specific social behaviors (soccer moms, for example) that suggested they would purchase particular vehicles or could be persuaded in certain ways. In addition to taking these differences into account, Ford's ad servers varied the messages depending on whether the person had or had not previously clicked on a Ford ad and on which Ford ads the person had been served previously. In addition, the agency presented ads to the friends of people on Facebook who seemed to be ideal targets insofar as they appeared to be in the market for particular types of cars. “On Facebook, you take friends into account,” said creative executive vice president Scott Lang when interviewed for this book. “So-and-so liked this; you will too. People find that creepy in the beginning, but … they slowly get used to it.”11
• Team Detroit was the first agency to use the Google Display Network of one million Web publishers to serve audiovisual commercials that matched the content of a publisher's site. It worked this way: The agency supplied Google with keywords that its research suggested would resonate with different types of individuals who might be inclined to buy automobiles. When a person went to a site containing one of the keywords, Google Display Network posted specific Ford display ads embedded with particular audiovisual commercials that were in turn designed to evoke those keywords. As AdWeek noted, “Depending on the context of the website, the most relevant Ford content is pulled and displayed in the ad unit, using video from the Ford YouTube channel and information from Ford websites thefordstory.com and fordvehicles.com.”12 So, for example, a site that discussed green issues highlighted Ford's hybrid models. A tech site might show a clip from Ford at the Consumer Electronics Show.
Andy Pratkin noted that these activities represent merely the tip of the iceberg. Ford, he said, has a wide-ranging capability to compile all of the demographic, social, and behavioral qualities it uses to define users’ profiles and then to present them with views of Ford products designed to match these profiles as the users move through the Web. He acknowledged that the capability certainly exists in the Google Content Network. Its wide-ranging collection of sites provides companies with the technology needed to track people with cookies and then serve them commercials based on their personal characteristics, their web-searching history, and the nature of the particular site on which the commercial is being served. According to Pratkin, as of mid-2010 Ford had not yet taken that step because the company was still testing its new capacity to selectively serve audiovisual commercials based on a site's context. He suggested that Ford already leverages these capabilities when serving text-and-picture ads on the Google Display Network and that commercials could easily be integrated into this approach.13
But the blending of conclusions about consumers’ online behavior with purchased information about them to tailor ads for them doesn't stop at particular brands. To the contrary, the business of discriminating digitally is leading many marketers to a small number of firms that sell a wide range of information about online customers and potential customers. These agencies, in turn, get their data from a small number of primary sources. From these cross-pollinating building blocks reputations are created as is, in turn, consumer discrimination in the form of narrowed options.
Acxiom, one of the top U.S. marketing-communication agencies, is a good place to begin exploring these building blocks and how they travel across advertisers. Established in 1969 to guide direct-mail firms with customer-relationship management, Acxiom has branched out to help clients “acquire, retain, and grow loyal (and profitable) relationships” in the digital media, especially via e-mail, banner ads, search engine optimization, website optimization, and cellular phones.14 The company's cross-advertiser reach is prodigious; Acxiom says its client list includes some if not most of the leading companies in all these areas: credit card issuing, retail banking, telecom/media, retail sales, automotive manufacturing, brokering, pharmaceutical manufacturing, life/health insurance, property and casualty, lodging, and gaming.15
Acxiom's use and analysis of customer data is central in serving these clients. Not only does Acxiom help its clients analyze their own information regarding particular customers, it offers a great deal of additional data about these same people in order to help the client better determine the value of specific individuals and tailor their sales approach accordingly. A major part of its role, the company repeats throughout its website, is to help clients decide which consumers are not worth targeting. This segment is called “contact suppression,” and it's based on making assessments of the “risk profile” of individuals for certain industries. Tom Mangan, the firm's senior vice president of consulting services, described Acxiom's contributions this way at a shareholders meeting:
Should I continue to target people who are in fact going to default on their loans? Or, take, for example, a telecom industry where they're constantly flipping cell phones. I want to understand those characteristics and to say, is that a customer I really want to target? …
So the opportunity here is, first, let us get an enterprise view of who your customer is… . That's the fundamental building block we have to have in place. Once we have that fundamental building block in place now, we can then enhance that with external information that we keep in our knowledge bases about every purchasing consumer across the country. And we can then use that to now start doing strategic segmentation, so we can break your customer base into each one of those strategic segments, from high value to low value.16
Acxiom's Consumer Data Products Catalog reflects the heart of this operation for the United States. The catalog promotes the agency's ability to sell “data enhancements” about individuals whom the client has identified by name and address or via some other unique convergence of information about them. Identification may have taken place through a website registration, a sweepstakes sign-up, or some other activity that explicitly or implicitly gave the marketer the right to use that identification for its own business purposes. The advertiser may want to know whether the person is worth pursuing or, if already a customer, how to best address him or her personally to achieve additional sales. The Acxiom catalog distributed in 2010 details what it offers about people in seventy-seven pages of single-spaced data codes, or “elements.” It claims it can help marketers by selling demographic data such as age, gender, marital status, race, ethnicity, address, and income “covering over 126 million households and approximately 190 million individuals.”17 But that's just the start. Noting that everything it offers conforms with U.S. privacy laws, the booklet sells quite particular constructions of the interests, lifestyles, and life circumstances of the population, including dozens of specifics about individuals’ online, catalog, and in-store purchases. For example, Acxiom places individuals into “interest buckets” that include such disparate categories as arts, celebrities, Christian families, and aviation. It divides people by their use of various credit cards and the recency, frequency, and dollar value of purchases from low-, mid-, and upscale catalogs. It sells health-related access to consumers whose purchases and self-reports indicate categories such as allergy-related, arthritis/mobility, cholesterol, diabetes, orthopedics, and senior needs. It can deliver individuals according to recent purchases organized into dozens of categories such as apparel/big and tall, donation/contribution, gardening, high-end appliances, and hunting. Acxiom assures marketers that “the freshness and accuracy of Buying Activity makes it … highly predictive in consumer models.”18
Acxiom statisticians combine demographics and various behaviors into seventy colorfully named and artfully described clusters arranged on an income spectrum, from the affluent (for example, “Summit Estates,” “Corporate Clout,” “Hard Chargers,” “Kids and Clout”) to the upper-middle income earners (“Soccer and SUVs,” “Acred Couples,” “Clubs and Causes”), middle-income earners (“Urban Tenants,” “Outward Bound”), lower-middle-income earners (“First Digs,” “Home Cooking”) and low-income earners (“Single City Strugglers,” “Mortgage Woes,” “On the Edge”). Here is where profiles get turned into reputations. To help its clients make sense of the data efficiently, Acxiom links many of the seventy clusters into twenty-one larger combinations, which it calls “LifeStage Groups.” The “Taking Hold” group, for example, brings together the “Married Sophisticates,” “Children First,” “Career Building,” and “Spouses and Houses” clusters under the conclusion that all are “fueled by wealth, as these clusters have already made it into the middle and upper-middle income brackets.”19
Probably the most useful frames for its clients, though, are packages of data that the firm customizes for particular industries. The agency's “enhancement data packages” bundle the kinds of details clients in particular industries typically use to make decisions about whether and how to pursue particular prospects. The number of elements about a person that each package includes varies by industry. Acxiom presents the auto industry with eighteen elements, for example, while auto insurance gets thirty-three; health care, forty; investments, thirty-one; travel and entertainment, thirty-six; and media, thirty-three. The company customizes its clusters for the insurance and financial-services industries—thirteen for the former and twelve for the latter. It also parses Hispanics in a separate cluster for companies seeking to sell to this fastest growing U.S. ethnic group.
These elements, clusters and grouping, are lenses through which many of the largest companies extend their understandings of their customers. They often include far more than the initial data that the client holds about a person. Speaking extemporaneously at a stockholders’ meeting, Mangan, the firm's senior vice president of consulting services, put it this way:
[W]hat we find today is [the utility of] breaking consumers into logical, strategic segments, understanding the attributes of those segments in terms of lifestyle, life stage, geo-demographic information, understanding the attitude of those particular segments, of the consumers and what they like to purchase and actually analyzing the behavioral information of the transaction data inside the company.
So you understand what are they in market for… . What is their product preference, how loyal they are to specific brands and then what channel do they like to be communicated. Once we have those together, now we can start doing effective targeting to that particular segment and be much more effective into what they want to buy.20
Use of Acxiom data between and across industries is intrinsic to the process. The work, company president John Meyer noted in 2008, involves “continually figuring out how we take that information but [use] it in new forms, different industries, different capabilities, so that we can open new markets for ourselves.” This repurposing of individuals’ data, he said, involves applying “what we have in solutions and take the best of those solutions and package them, product-ize them, put a little box around them and say what value does a [client] see of being associated with this.”21 Corporate marketing strategies will vary. But in the large number of cases where an overlap might exist for current or future consumers for cars, financial firms, and insurance companies, Acxiom bases its presentations to advertisers within and across industries on the same data about individuals. It reports, for example, on the individuals’ online shopping amounts, income levels, interest areas, discretionary-income score, and vehicle propensity. And it even brings those data together to tell the same stories about individuals to all clients. Think of a person Acxiom positions as an “outward bound,” “married sophisticate” who is a “chiphead” (the company's catalog doesn't define this appellation) with a MasterCard and a “diabetic focus.” That kind of profile—and the story about the person that is built into it—often travels far.
As if to illustrate the point, Acxiom turns out press releases touting its willingness to sell data about the same individuals to all comers in the auto industry. A 2009 self-promotion, for example, offered to “reveal consumer-centric insights across more than 130 million U.S. households [about] demographic and life stage portraits, shopping behaviors, attitudes, intentions and propensities toward vehicle purchase and ownership for 2009.”22 The company was selling its pinpointing of particular households and consumers who fit the various profiles to any interested advertiser. Buyers might use the data differently; Ford might decide that the individuals in certain of Acxiom's propensity segments ought to get its ads and attention, while Cadillac might prefer only some of those in addition to other prospects Ford didn't pick. Still, both would be making decisions about how to address Americans based on the basic stories that Acxiom has told about them.
What takes place within Acxiom also occurs in numerous other firms that provide data to media buyers for targeting. In presentations to clients during 2010, executives of the media-buying agency MediaVest listed Acxiom as one of five companies that it was negotiating with for targeting consumers on behalf of its clients, as well as noting thirteen other data-centric firms with which it already had such arrangements. Acxiom's expertise, according to MediaVest, is in providing “very granular data around demographics and psychographics within households, including segments such as education level, cars owned, and life stage.” The other firms—for example, BlueKai, eXelate, Media6Degrees, and Mindset Marketing (now called Medicx)—have reputations for delivering particular types of prospects to marketers via cookies for online advertising campaigns.
Unlike Acxiom, which typically offers people who can be identified by name, these firms generally offer individuals who are anonymous because either they are tracked by cookies that are not tied to registration material, or the firms selling the cookies have stripped out personally identifiable information as a result of the privacy policy under which it was collected. Either way, marketers know that a certain percentage of those reached anonymously can be seduced into revealing who they are when they are offered a sweepstakes entry or information about a product sent to their e-mail address. The information gathered that way can then be used to purchase more information about the individuals from companies such as Acxiom so that inter-actions around products can be as “relevant” as possible. But even when marketers don't know the names of the people associated with the cookies, the data they hold can still be extremely valuable for making judgments about their interests and the most persuasive strategies to advertise to them.
Companies that sell access to individuals via networks or exchanges typically specialize in profiling people's lifestyles or buying intentions, which they determine by placing cookies on people's computers to track their visits to particular sites. The companies then analyze what those visits generally mean about an individual's interests. Not surprisingly, each company contends that the information it collects and analyzes is superior to that of its competitors. Some cookie providers insist, for example, that the act of clicking on products while visiting websites is the key to understanding a buyer's intent and therefore cookies that report such “in-market” behavior provide the best insight into the products that should be advertised to particular people. As discussed in Chapter 3, BlueKai and eXelate specialize in this domain. An eXelate “intent” segment called “auto buyers,” for example, includes planting cookies on computers of individuals who have visited “leading auto research and auto dealer lead generation sites” and who, eXelate concludes, are “in-market car buyers that are actively searching to purchase a vehicle.” For its sixteen “shoppers” categories (one of which is “babies and kids”), eXelate uses “comScore Top 10 shopping sites and vertical sites that track product interest and drive specific purchasing behavior.” On these sites it tags people with cookies when they are “looking for pricing, ratings, and purchase information on specific products [for example, cribs] as well as enthusiasts commenting on and shopping for [the] niche-interest [that is, the baby and kids-related]-items.” It then sells to media buyers the right to reach people via those cookies.23
Other cookie providers disagree that in-market behavior is necessary for understanding what a person will buy. They say that more can be learned from the terms people use on search engines such as Google and Yahoo!. A firm called Magnetic contends that “the integration of leading-edge search technology and advertising is the wave of the future.”24 The company inserts a Web beacon on sites serving Google and Yahoo! search ads, and the beacon creates a cookie that records the search words that led the person to the site. Magnetic then offers marketers access to those cookies as a way to “retarget” ads to individuals virtually anywhere on the Web based on the interest they expressed via the search term(s) they used.25
Yet another approach to finding people who are likely to buy certain products involves collecting data about individuals’ off-line areas of interest and then providing this information to marketers so they can advertise to them online. Two areas of particular concern are health care and financial services; some advertisers will pay considerably for the chance to get the business of Americans who spend a lot of money in one or both of these areas. Medicx Media Solutions is one such “vertical” data provider. The company contends that its “Geo-Medical Targeting” “minimizes waste” by helping advertisers narrow the locations of individuals with specific “diagnosed medical conditions, prescribed treatments, both Rx and OTC, health styles, type of insurance coverage, co-pays, and medication compliance or consumption measures.”26 To facilitate its licensing of this information, Medicx obtains medical and pharmacy insurance claims data for tens of millions of Americans from third-party data providers that purportedly strip out personally identifiable information irreversibly in conformance with regulations set by the 1996 Health Insurance Portability and Accountability Act (HIPAA). Even though Medicx supposedly cannot tie the data to particular individuals, it has developed statistical techniques (which the company is attempting to patent) that it says can connect the medical and pharmaceutical findings to the U.S. Postal Service's ZIP+4 system. That scheme breaks down ZIP codes into areas as precise as a city block or an apartment complex, or even into a portion of a street comprising as few as three to eight homes. To enable advertisers to reach these patients online, Medicx licenses millions of cookies with ZIP+4 data and then serves its clients’ display ads to individuals in the targeted ZIP+4 areas. The company states that this process has “turned the art of reaching audiences by medical condition across virtually any digital medium into a measurable and accurate science.”27 The people receiving the ads about specific medical concerns would have no clue as to how they were targeted.28
The companies discussed here are just a few of the choices ad agencies have when they are trying to decide whom to research and target. A particularly popular approach involves mapping interpersonal relationships and predicting the value of those relationships, as we will see when we explore social-media advertising in Chapter 6. Clearly, the notions that advertisers would have about their targets would vary depending on the data firm they choose, and perhaps even depending on the segments that they use for targeting. Advertisers also have the ability to adapt to the particular responses that consumers make to their offers. A training manual from Omniture, a company that helps firms make these modifications, states that “dynamic models automatically adjust to changes in propensity over time, both for individual visitors and the population as a whole. In other words [the Omniture system] constantly refines its understanding of ‘who you are'/'your preferences.’ And then automatically (i.e. in real time) serves you different content as it learns more.”29
The transient nature of cookies is another factor that encourages variability in the ways individuals’ profiles are constructed. A segment of the population routinely erases cookies; newer browsers sometimes help them do that. Moreover, some data providers consider cookies that identify people who are seeking particular products to be useless or of little use after a couple of weeks since by that time those people are likely to have made their particular purchase and have moved on to other things. But data executives largely do not seem to be troubled by the temporary nature of cookies. They simply note that a new cookie will be created in short order when the person returns to one of the many sites that their companies use for placing cookies on computers.
These chances for variations and change in peoples’ profiles contribute to the varying approaches that advertisers use to market to an individual, and indeed sometimes perceptions of a person's interests and value may even be at odds among advertisers. Nevertheless, the evolving mechanisms for generating longer-lasting profiles and the increasingly broader sharing of them also need to be noted. For one thing, companies that want ongoing knowledge of visitors to their websites can certainly find ways to get around cookie deletion. For example, some firms have turned their third-party cookies—the kind that ride on ads and that people typically delete—into first-party cookies.30 These are the “good” cookies that help websites remember who you are—for instance, based on the preferences you set on previous visits to a website, cookies will communicate to the site how you personalized the appearance of its home page. By hiding among the good cookies, the bad ones can stay active longer. Storing a computer's unique internet protocol (IP) address to detect its return to a site is an alternative to cookies. Other approaches conceal cookielike identifiers that are difficult to detect and erase in places on a computer's hard drive. A less-direct approach to storing and preserving a person's data is to connect anonymous cookie-linked data to legally collected, personally identifiable data—for example, by encouraging a person using a computer on which a cookie has been placed to fill out a registration form or sweepstake entry that requires a name or e-mail address. Databases can hold on to this and other combined information even after the individual's cookies are lost. The stored data can be reconnected to the person when he or she again identifies herself and that identification is linked to a new cookie. Such a system violates no existing privacy law, and, judging by my conversations with industry personnel, it's an activity that is growing.
Just as significant as the persistence of people's profiles is the spread of consistent information about individuals across the digital universe. Despite what seems to be a segmented online world, the core elements that data packagers use for distinguishing and classifying consumers and discriminating among them are surprisingly extensive. Virtually all the companies in the business of describing consumers for advertisers rely on the same small number of providers. The three major agencies that keep tabs on people's creditworthiness—Experian, Equifax, and TransUnion—are particularly important sources of information. Experian, for example, offers marketers more than 1,200 pieces of lifestyle data on more than forty-seven million households; address verification; ways to reach individuals “at the precise time they are shopping for credit” and more.31 Reed Elsevier, another big information provider, positions itself as helping companies manage risks through such services as “Know Your Customer (KYC),” identity verification, fraud analytics, due diligence, and credit risk assessment solutions.32 Most of the data packagers, in turn, start with public records such as U.S. Census data, county records, tax assessments, and telephone directories. Once data-exchange firms tie previously anonymous cookies to personal identities, they still need to conform to the anonymity requirements of various privacy policies before selling them on networks and exchanges, so they append the details about the individuals to the cookies and then make them anonymous again. The demographic and lifestyle categories of a person's profile may well appear across many of the cookies that different advertisers use, and these classifications often fall within segments that marketers and agencies such as Acxiom assign to consumers and that broadly guide decisions about whether and how to pursue specific individuals. These assessments include whether an individual should be considered a desirable or an undesirable customer, whether a person should receive an ad for a particular product, what the ad should look like, how much the product should cost, and even what the ad copy should say.
An area where these sorts of social discrimination have already become part of everyday thinking is the digital discount-coupon business. Despite the huge popularity of paper coupons, online coupons finally began to take off in the late 2000s. Gian Fulgoni, the head of the comScore internet ratings firm, was surprised when he noticed increasing enthusiasm for online discounts in October 2008. “Coupons had never been a big factor online the way they are offline. This is something new,” he told the New York Times.33 ComScore estimated that twenty-seven million people visited a coupon site during that month, up 33 percent from a year earlier. Coupon distributors assumed that the rise in digital coupon searches was a result of the dire economic downturn that was now facing the nation. The use of coupons clipped from newspapers and other places were enjoying a boost, too, after several years of nominal growth. The trade magazine Progressive Grocer reported in late 2009 that 90 percent of all coupons were still being distributed through freestanding inserts (FSIs) in newspapers. The real news, according to the magazine, was in the redemption rates. Only 50 percent of the coupons being redeemed were those from FSIs. The rest, and the fastest growth, was coming from online coupons delivered through the internet and mobile phones, and at booths in stores. According to the coupon-processing company Inmar, redemption increased 263 percent to about fifty million coupons in 2009. “Consumers can simply print the coupons at home or at a kiosk in the store, or download them to their cell phones, and redeem them at the checkout,” Progressive Grocer noted.34
In 2008 many companies were irked by the spike in coupon use, because the technology made it easier for people who found coupons online to use them. “When the coupons get wider exposure,” the Times noted, “retailers lose control potentially costing them more money than expected.”35 In spring 2008 Harry & David, a high-end seller of gift and fruit baskets, threatened legal action against the RetailMeNot coupon site for publishing its discounts. The online and catalog retailer Sierra Trading Post in 2006 had to sell $300,000 worth of merchandise at a very low margin because someone posted its unlimited-use catalog coupon code on the Web, and the company noted that it was looking for ways to control its allocation of discount offers. Adopting the apocalyptic, consumer-power rhetoric of the day, Fulgoni said that online coupons have “taken pricing power away from retailers and given it to the consumers, because the consumer is totally up to speed on what the prices are.”36
More recently, a New York Times article took a wholly different approach in describing the coupon industry's efforts to create technologies to regain leverage over shoppers. The article's first line sums up its theme: “For decades, shoppers have taken advantage of coupons. Now, the coupons are taking advantage of the shoppers.”37 Part of this new advantage—a development the article doesn't discuss—are the methods that digital-coupon sites are using to protect a retailer's revenues by limiting the number of times a digital coupon can be used—“distribution control” is the industry phrase for it. The article focused on a related development in digital coupons: personalization, noting that the bar codes of coupons printed from the internet or sent to mobile phones “can be loaded with a startling amount of data, including identification about the customer, Internet address, Facebook page information and even the search terms the customer used to find the coupon in the first place.”38 All that information, the article continues, would follow the customer into the physical mall, which is where most consumers still make their retail purchases. Sometimes the coupon-holding shoppers are anonymous, other times they aren't. The co-founder of RevTrax, a company that facilitates the personalization process, gives an example of how it works with Facebook for the Filene's Basement clothing store:
“When someone joins a fan club, the user's Facebook ID becomes visible to the merchandiser,” Jonathan Treiber, RevTrax's co-founder, said. “We take that and embed it in a bar code or promotion code.”
“When the consumer redeems the offer in store, we can track it back, in this case … to the actual Facebook user ID that was signing up,” he said. Although Facebook does not signal that Amy Smith responded to a given ad, Filene's could look up the user ID connected to the coupon and “do some more manual-type research—you could easily see your sex, your location and what you're interested in,” Mr. Treiber said.39
The article goes on to specify that, according to an executive from Filene's digital agency, “Filene's [does] not do this at the moment.”40 But even if the retailer doesn't link the Facebook ID to the demographics in its database of store customers, it could still draw important conclusions about who would respond to what discount offers—which is the whole point of using RevTrax. Categorizing people simply by their registration answers on the company's Facebook “fan site”—a spot where people can sign up for more discounts and interact with others who share their interest in the company's products—would help the retailer maximize revenues from different current customers. It could also help the firm's media buyers decide on how best to pursue “lookalikes”—people whose profiles resemble individuals with certain amounts or types of Filene's Basement purchases. This group would receive coupons via display ads or e-mail with the media buyers’ unique codes on them, while people with different profiles wouldn't get the coupons.
Filene's evident desire to distance itself from these activities reflects a discomfort by some retailers about talking publicly about their involvement in using data the customer hasn't knowingly presented to the retailer. This uneasiness is also reflected at the popular Coupons.com, which proclaims, “We enable digital coupon programs for the nation's top brands and advertisers.”41 The home page of Coupons.com invites people to join and shows examples of the sites’ offers. It suggests a person might receive discounts for products based on where he or she lives; a “local” tab seems to be guided by the IP address, and the company asks for the registrant's ZIP code. Nowhere on the home page or in the description of the firm's activities in “About Coupons.com” is there a suggestion that offers will vary based on a person's social categories and activities on the site.
The only mention of this is buried in the firm's privacy policy. Several paragraphs after a preamble attesting to “the importance of protecting the privacy of all information provided to us [as well as] information we collect” in “our relationship with our users and consumers who use any of our websites, co-branded Microsites, services, or applications owned or provided by us (the ‘Sites') (collectively or individually, ‘Consumers'),” the policy states that “when you visit the Sites, you may provide to us what is generally called ‘personally identifiable’ information (such as your name, email address, postal mailing address, and home/mobile telephone number, etc.) if you print a coupon from or otherwise participate in activities offered on the Sites. In addition, you may have the option of providing us with certain demographic information about yourself or your household (such as age, gender, household compensation, interests, zip code, and state).” Buried within the next section (“The Way We Use Information”) the policy explicitly states that the firm uses cookie data to “provide custom, personalized coupon promotions, advertisements, content, and information.” By way of explanation of its right to personalize based on information requested directly from consumers, the policy states that it uses such information “to understand the usage trends and preferences [as well as] to improve the way the Sites work and look.”42
The company separates cookie-based findings from information consumers give the site directly in order to comply with its statement that “we do not link the information we store in cookies to any personally identifiable information you submit while on the Sites.” Yet this statement must surely puzzle a visitor who continues reading, paragraphs later, that the company may well give its advertisers the ability to customize coupons from data yielded through linking cookie and personally identifiable information. The policy says, “We disclose both personally identifiable and automatically collected [cookie-based] information to our clients … to process such information on our behalf,” and it adds, “We disclose automatically collected data (such as coupon print and redeem activity) to our Clients and third-party ad servers and advertisers. These third parties may match this data with information that they have previously collected under their own privacy policies, which you should consult on a regular basis.”43
Clearly, Coupons.com acknowledges in this section only the possibility of tailored coupons by its sites and its advertisers while not saying what it actually does, who really does it, how extensively, and based on what information. Yet such coupon tailoring is an activity that happens frequently in various digital formats. In an interview for this book, Jonathan Treiber of RevTrax described a basic retargeting routine that inserts coupons on display ads, using as an example shopping for power tools at the Sears website. He said, “You don't have a Sears account. You go to check out and you abandon the shopping cart. A banner ad will show up on another site, saying ‘Come back to Sears and we'll give you 20 percent off your power tools.’”44
Don Batsford, Jr., a partner at the digital agency 31 Media, agreed that data-driven coupon customization has become the norm. His clients are interested in direct response, and he works “at the bottom of the purchase funnel”—the point at which a customer is ready to be persuaded to buy one or another product brand. Batsford prefers to skip social networks and general display and focus instead on paid search, which involves an advertiser auction for words that individuals might type into the box of a search engine such as Google, Bing, or Yahoo! (which is powered by Bing). The highest bidder isn't necessarily the winner. The search engines use secret formulas to evaluate the relevancy of a proposed ad to the keywords the advertiser is bidding. The tighter the connection between the two, the more likely that advertiser will win the bid. Search engines also consider past successes of the firm's ads in getting people to click on them; a history of clicked ads weighs in a bidder's favor.45
Ads for the winning bids show up to one side of the organic search results as well as above them. To Batsford, paid search is “Zeus” because, he says, keywords indicate people's interests at the moment of searching, as opposed to tracking past behavior on websites or comments on Facebook. “They confess to the search engine [and] are very open about what they want.” Because of the ads competing on the page, “you have to have the neon sign to get people to pay attention, so you need the coupon.” Batsford uses data from the search engines as well as the RevTrax system to determine the characteristics of the best responders so that he can know whom he will offer the best deals in the future, as well as to correlate keywords with spending habits.46
The hottest, and perhaps thorniest, issue regarding coupon tailoring involves customizing the prices. “This is really the holy grail in a sense, pricing to the individual,” noted Willard Bishop, a retail consultant in suburban Chicago who focuses primarily on supermarkets.47 Batsford agreed, but he insisted that certain forms of coupon discrimination online could backfire. Sending people offers for different products based on keywords and background characteristics is fine, he said, but varying the discount amount for the same product can work only if a company carries it out very carefully. His concern is not about ethics; using data in this way falls squarely within privacy regulations, he said. Rather, he said, building distinctions based on demographics, search, or behavior won't work if simply presented to different consumers on the same website. People would find out about the various discount levels and figure out ways to game the system and always get the biggest reduction. He said he did feel comfortable giving new customers higher discounts than those given to existing customers because the website would be able to identify the existing customer and not give that person the higher discount. Similarly, “friends and family club” members for some of his clients could realistically be separated out for special treatment.
This idea of selectively presenting differently priced offers while precluding nontargeted consumers from taking advantage of them seems to be the evolving norm in the business. Jonathan Treiber of RevTrax noted that although his company doesn't approve of the ethics of evaluating people based on their movements through the Web, he knows of many firms—including his clients—that carry it out to decide who should get their coupons. He argued, though, that going further and giving chosen people different discount prices without tying it specifically to them won't work because “internet information gets shared rampantly,” and the coupon deliverer cannot be absolutely sure a discount coupon will be redeemed by the intended shopper. A company can be assured that a specific consumer will get a coupon, Treiber continued, if its site has log-ins—or cookies that do the signing-in for the customer. In this way, “Coupons can only be redeemed by Joe. If you get the coupon, you can only get it once,” he said. “That may piss other people off; that's neither here nor there. People get different coupons. The client is not concerned because it's clear as day that it's exclusive to Joe.”48
Other Web coupon practitioners agree. They say the aim of price and product personalization is often to lure a potential or actual customer who has particular buying patterns by giving him or her a particularly attractive (and often low-margin) offer that will ignite or reinforce a lucrative relationship. From their perspective, the trick to achieving successful product-and-price personalization involves exploiting as much data as possible about consumers without violating privacy regulations while also making sure that the separation of consumers is secure so that unintended people won't be offered these precious deals. Omniture is one of several firms that provides websites and e-mail servers with the capability to carry out these activities. The company's training manual says that, for visitors arriving at a client's site from an e-mail campaign, Omniture will “make sure that only those visitors are served promotions that align with the special offer they were given in the email, thus ensuring consistent messaging throughout your site.”49
If any company has taken this data-driven online separation of targets and waste to its logical conclusion, Next Jump has. The company has been unknown to all but close observers of retailing even though it runs employee discount and reward programs on behalf of ninety thousand corporations, organizations, and clublike associations (often called affinity groups) that reach more than one hundred million consumers. Its low profile reflects Next Jump's desire to remain in the background in order to broker or facilitate deals. On its website it depicts itself as giving “over 28,000 merchant partners, both retailers and manufacturers,” the technology “to create the most targeted, cost-effective and measurable campaigns to reach more than 100 million users.”50 The capability it offers prospective clients on its website is certainly familiar in the digital age: “Simply created targeted ads that are promoted by free emails. Gain insights to your marketing performance to improve future campaigns. Target your most profitable customers.” Next Jump's ability to target the most profitable customers depends on a trade secret that, according to founder and CEO Charlie Kim, augers a new future for advertising. The business is based on the proposition that the best way to advertise is to present lower prices specifically to individuals whom statistical calculations suggest are the best customers to reach with a deal. According to Kim, Next Jump's conversion rate on product offers is 11 to 1, compared to 1000 to 1 or worse for typical internet ad-conversion rates. But not every advertiser can get the data to target individuals so carefully. What Next Jump has done is insinuate itself directly into the fabrics of Fortune 500 companies and other larger firms by running the online purchasing programs those firms set up as perks for their employees. Those firms pay Next Jump for its work on a per-employee basis, but the company seems to derive a lot more income from the fees merchants pay every time someone buys a product. Next Jump therefore has a vested interest in increasing the purchases. Because Next Jump operates discount programs for roughly one-third of all U.S. corporate employees, “it is considered a non-traditional benefits provider and gets updated weekly on the employment status of 30 million workers (who also happen to be consumers). It gets part of the employee record, including things like name, address, employment status, home and work address, marital status, and sometimes even job title or salary grade.”51 To that information and inferences about it (such as salary grade), Next Jump adds transactional data from the firms it deals with as well as from credit companies. It then overlays all that with information it collects through the firms’ human relations departments and employees to determine not only what people want but what they will pay for. With all that knowledge, Next Jump can create predictions of who should get what product e-mail offer at what price. It also generates what it terms a “UserRank” score for every employee based on how many purchases a person has made and how much he or she has spent.
“Shopping and advertising has always been the same to us,” Kim noted in 2010. His goal is to separate consumers to the extent that the conversion rate (which in 2008 was 100 to 1) will get down to 3 to 1. He sees his mission as a statistical challenge. Of his 225 employees, 150 are engineers. One industry analyst noted, “If Kim keeps perfecting his shopping algorithms, you may never shop the same way again—and you won't even know that you are doing anything differently.”52
The mining of data, its circulation, and its use for personalization doesn't stop with commercial messages and discount offers. As the next chapter shows, publishers, eager for advertising revenues, are likewise deeply involved in trying to carry out targeted personalization of their content. And in the process they help the advertisers keep what they're doing hidden from public view.