Oh, Behave!
"I've found my perfect customers, and I know what they want."
What marketer hasn't woken up having dreamt of being able to say those words? And yet even the most talented in the field have been more apt to wake up in a cold sweat instead, despite tools at their disposal such as analytics and segmentation, custom-packaged direct mail and strategic keywords. In the near future, though, marketers may awaken to a brand new day: What better way to know what consumers want than to watch how those consumers actually behave?
The concept of behavioral targeting (BT) has been around since the late '90s, disappearing for a few years after the dot-com bomb, according to Andy Chen, vice president of digital solutions for Viacom Brand Solutions International. At the time, Chen says, companies such as Engage (Flycast) and DoubleClick only "dabbled" in BT technology. Today, behavioral targeting is acting out -- and experiencing a comeback.
Technology has enabled marketers to tap into behavioral targeting, allowing them to watch how individuals are interacting with the online experience and to predict, in real time, the best message to launch. For all marketers, having the ability to respond "in the moment" has become a crucial strategic advantage. "Companies don't want to offer consumers products and services they don't want," says Richard Howe, chief marketing and strategy officer for data management vendor Acxiom. In the same respect, he says, "consumers are tired of getting stuff that has no relation to what they're interested in." What consumers want isn't a secret anymore -- and technology is empowering marketers to deliver.
With advanced technology and a wealth of customer data, the growing importance of BT is evident by the burst of acquisitions in the space in 2007, most notably Google's move in April to acquire online advertising pioneer DoubleClick for $3.1 billion in cash. If the deal goes through -- regulatory approval was still pending at press time -- Google gains access to an incredible store of information, making it even more at the vanguard of behavioral targeting and advertising. As one industry pundit put it, "The company could know more about Web surfers than they know about themselves." In the aftermath of Google's DoubleClick announcement, other acquisitions included
Yahoo! snagging BlueLithium for $300 million and AOL buying Tacoda for an undisclosed price in September. These moves and others have led observers to suggest that the next golden age of behavioral targeting is around the corner.
Who's With Me?
How
far around the corner remains to be seen. Much of the industry is still lagging behind the early adopters, according to Michael Greene, research associate at JupiterResearch. In a survey of 277 online advertisers, only 16 percent reported using behavioral targeting. Whether that sounds like a good or bad result, what's interesting is that it's unchanged from the results of a similar 2004 study, when "16 percent" represented a huge jump. That year, Nate Elliot, associate analyst at JupiterResearch, noted that the share of advertisers using BT went from 10 percent in 2003 to 16 percent -- a 60 percent increase. Elliot had a presciently pessimistic view, however: He was quoted at the time as saying that, with the hype and explosive growth, the area would continue to flourish, but "it'll never be a dominant force in the market." As big ad servers get in the game, he added, "the technology will be commoditized," hence making it difficult for others to rely solely on BT.
Three years later, that's turned out to be a solid prediction. And the future may be no better: Only 19 percent of marketers anticipate using the technology in the next 12 months. "Behavioral targeting has yet to gain broad usage, but it is certainly on the mind of a lot of online advertisers," Greene says. As more marketers embark on this new wave, they will hopefully overcome what Greene reports to be their most persistent plague -- finding the right audience.
Meanwhile, for those who have successfully implemented behavioral targeting capabilities, the payoff has been tremendous. Tim Vanderhook, chief executive officer of online media network Specific Media, claims that compared to targeting by category (e.g., "sports fans"), or by mere clickthrough rates, behavioral targeting has doubled marketing's efficiency. Acxiom's Howe says his firm has seen two to three times the return; one Acxiom user, Andrea Palmer, manager of interactive services at Baltimore-based advertising agency Siquis, estimates 30 percent to 50 percent lower costs of advertising for her client, Spirit Airlines.
Behave Yourself
And there are other drawbacks to having a relatively young presence in the industry: "One of the most important things about behavioral targeting right now is that it's really ill-defined," Greene says, especially in terms of execution. (See sidebar "How Do You Define Behavioral Targeting?" below.)
Much BT is highly dependent on the data that consumers provide through their Web activity. Historically, explains Specific Media's Vanderhook, companies that delivered adware -- advertising-supported software -- were the ones most often executing behavioral targeting. Every site a consumer visited was logged, triggering pop-up advertisements bundled within the adware. Not surprisingly, consumers became increasingly frustrated being interrupted by pop-up ads, marking their inevitable demise as a medium.
Today, marketers have access to both customer identity (IP address) and customer personality (online behavior). Now, BT often relies on HTTP cookies -- small pieces of data used to track and maintain the online actions of a user, such as sites visited and even items stored in an electronic shopping cart. "A company is only as good at behavioral targeting as the amount of data [it has] access to," Vanderhook observes. So while the cookie history is in place, potentially relevant ads can be delivered, but once it is erased -- or if the customer tries to access the same site from a different computer -- "it's like a brand new person," he says.
Tools like cookies, however, have been subject to debate because of privacy concerns. To reassure wary marketers and their consumers, Howe explains that the concept of targeting "individuals" is actually a misnomer. Although behavioral targeting is significantly more relevant, it's not quite the Big Brother consumers fear it is. Behaviors are entirely based on what is registered through the server and, at most, combined with household demographic information obtained through public records. Currently, Vanderhook says, "I could never make a [random] ad that says, '
Name, click here for travel ads.'" But as databases get larger and more personal information is stored on the Web, security issues may become more of a concern, he says -- perhaps leading regulators to try to restrict the industry in the name of individual privacy. Draconian reforms could hamstring behavioral efforts and innovations.
Another pitfall of targeting based on IP information is the discrepancy of shared computers. "We can't identify if someone gets up and another sits down," says Palmer, the manager at Siquis. Still, she believes that public computers are a minor concern for most marketers; in fact, Palmer says, "the times are few and far between when that many people are swapping computers to the point where the advertising might be skewed."
Catch Me If You Can
The more widely recognized tactic of behavioral targeting is tracking individual behavior and marketing based on a "rule set," or serving ads based on the recency and frequency of a specific search topic. When the software detects high viewing rates of travel sites, for example, this presumably indicates that a consumer is nearing the end of her researching cycle and entering into buying mode. At this point, Vanderhook says, that consumer will be hit with more travel ads offering deals related to her previous search history. Still, he notes, even this approach has differing tactics: targeting consumers based on Web site visited, keyword searched, keywords read within content, or actions on your own Web site (e.g., "homepage viewers," "product information viewers," or "shopping cart abandoned viewers").
Not everyone agrees this is the best method--after all, customers are rarely ever just interested in one thing, says Jack Jia, chief executive officer of Baynote, a provider of what the company calls on-demand recommendation technology. "Clicks are more of a function," he adds. "They're not a good indicator of what's useful.... That's why the number-one-used button [on a Web site] is the 'back' button." Jia argues that tracking the individual based on historical behavior stereotypes the consumer into being an insular person with singular wants. In response, Vanderhook says that clicks do provide useful insight when monitored for frequency over a short period of time. "You're not going to accidentally click eight times," he says.
Though Jia certainly has a point, when it comes to tracking across sites, targeting based on recency and frequency seems to be the best bet for today's marketers. "[A consumer's] historical Web-surfing behavior is absolutely critical," Vanderhook says.
Perhaps, then, one of the main reasons so few marketers are jumping on the behavioral-targeting bandwagon is because BT often requires more behavioral data than is currently available, according to JupiterResearch's Greene. Products or services that require longer purchase cycles and deeper research are able to satiate that prerequisite. As Vanderhook mentions, shopping for cars takes an average of six weeks -- plenty of time for marketers to dish out campaigns.
To target customers with multiple existing interests, Vanderhook explains that ads will only appear once online behavior in a particular category picks up. In the case of discount carrier Spirit Airlines, a consumer may select a trip from New York to Fort Lauderdale but exit the site before purchasing. If she lands on another travel site within the same advertising network, she will be presented with a traditional banner ad from Spirit reading, for example, "Low fares from New York City," or "Low fares to Fort Lauderdale." When consumers do leave a site before completing a transaction, Vanderhook, for one, is not afraid to make sure the consumer knows what benefits your product can offer over the competitor's. "You hit them at the next site that they're at, the next site, then the next site -- with different messaging to make sure they're truly understanding what you're offering," he says.
As much as behavioral targeting focuses on delivering "in the moment," marketers are advised to know when to throw in the towel. Palmer has frequency caps on her campaigns: After a certain number of campaigns or after a certain time period, ads will no longer be delivered to that user. Persistence has its limits and realistically, "we don't want to waste any more advertisement on that person," Palmer says.
For marketers who are just beginning to test the waters, Vanderhook suggests first experimenting with "retargeting." More straightforward, cost-effective, and less intrusive, Vanderhook believes "it's the single most powerful form of targeting in terms of turning visitors into customers." When an individual leaves your site and then later appears on a general, non-brand-specific site, marketers can retarget by hitting her with the ads for the specific product again, even outside their ad network.
Not So Fast...
Behavioral targeting is further constrained by practical scalability as marketers strive for deeper granularity. Greene warns that attempting to reach every individual with highly targeted campaigns will be very difficult, very time consuming, and very costly. Experts advise partnering with advertising networks -- a group of Web sites that want to host ads with advertisers that want to launch ads. According to Vanderhook, ad networks can "provide the scale and efficiencies" unattainable with site-by-site implementations.
To judge whether or not your marketing team is prepared to dive into behavioral targeting, Greene proposes three broad, but essential, questions marketers must have the answers to before taking the leap:
1) Can we find enough information to verify that this person is using legitimate search, and is this type of behavior in accord with our marketing objectives?;
2) How can we find the exact types of behaviors we are looking for?; and
3) How can we spread our message based on these behaviors and make sure it gets to the right consumer?
Simply put, companies need to understand themselves and what their own campaign goals are before expecting behavioral targeting to save the day, Greene says.
At the same time, he adds, behavioral targeting doesn't simply deliver on expectations. Marketers should be willing to learn from new and unexpected insights as a result of the technology. During her campaign for Spirit Airlines, Palmer was surprised that the group she logically assumed was the airline's most popular customer -- large families traveling with multiple people--wasn't. In actuality, consumers were more likely to be married couples without children. Incorporating this knowledge, Palmer says, "will help us open the door to additional advertising opportunities in the future."
On the Home Turf
Within the home site, behavioral targeting can work in a slightly different way. An individual is not fed ads based on what the marketer detects to be a single defining identity. Once consumers are on a site, behavioral targeting shifts from displaying relevant ads to relevant content. "Intent is one thing. Interest is one thing. But whether or not the content or product behind [a click] is useful is a completely different thing," Jia says. He believes that individual behavior is often insufficient to determine if the results are really serving the needs of the user. Baynote's technology is based on "the wisdom of crowds" (also the title of a book by
New Yorker columnist James Surowiecki). Using concepts derived from social psychology, Jia believes a "crowd" (at least seven people who act in the same pattern) can better judge the value of a particular piece of content to that specific group than an expert.
Jia argues that past efforts to target based on personalization within sites have not worked because marketers were fixated on "rules-based personalization engines," and, as such, blamed failure to the lack of data. While data is important, he says, "The issue is not about having
more data. You can't personalize
me, because there's too much of
me," he adds, noting that it depends on the context of the behavior displayed. Product recommendations on the home site differ from those on external sites because even with all the data in the world, you'll likely only end up displaying the most popular, not the most relevant.
By understanding the principles of social psychology, Jia recognized that, as in real life, individuals behave alike as a result of their context. Users, he says, are demonstrating what's useful, or not useful, to them based on their actions on the site -- spending time on a piece of content, interacting with it. "They're voting with their feet," he says. Essentially, marketers can map the behaviors of like-minded users to determine what the most relevant content should be.
Businesses are still reluctant to let go of what Jia refers to as the "expert mentality." Believing they know what the customer wants, marketers fail to reap the benefits of allowing their customers to organize their own preferences. As an example, Jia refers to how a search for the word "hammer" on HomeDepot.com does not bring up what customers would immediately want when searching for "hammer." Instead, the first page is dominated by power drills that have the word "hammer" in the product name.
US Appliance, a Baynote client, recognized a similar inconvenience on its Web site. When customers search for "stove," results often lead to other appliances because the technical name for a stove is "cooktop." After implementing Baynote, the company's Web site now has a separate "recommendations for item" box at the top, which reflects the course other consumers found most applicable when searching for similar products.
The same was true at Education.com, a publisher of scholarly works focused on child education. "We really wanted our users to become part of our site," says Todd Schwartz, vice president of marketing for the company, where he helps parents access the most relevant information as conveniently as possible. "Strict search wasn't doing it for us," he says. "We want to make sure that when you're on that page...you're surrounded with content that is consistent with that experience." To achieve that goal, Education.com rejected basic search technology and instead relied on the research paths forged by users who have given their "manual approval," he says.
This technology is present on other Web sites, such as Amazon.com with its "Customers Who Bought This Item Also Bought..." function, and, similarly, Netflix's "Movies For You." (See
"Playing At the Speed of the New," December 2007, for more about Netflix's successes.) In fact, Netflix is offering $1 million to those who can "substantially improve the accuracy of [preferred-movie] predictions," based on selections made and self-identified preferences. (More specifically, the company is looking for technology that can achieve recommendations 10 percent better than its current software, Cinematch.) On each of these sites, the technology creates a highly relevant experience, which will in turn increase conversion rates.
What's Next?
Although behavioral targeting has yet to achieve mainstream adoption, experts have no doubt the rest of the industry isn't far behind. "Much like direct mail in the [1970s]...people started off slow -- didn't know what they were doing, didn't know how to target," Howe recollects. Eventually, he says, people got progressively better, and he predicts the same pattern will unfold for behavioral targeting as marketers apply offline lessons to the online world.
As BT reaches higher levels of sophistication, Vanderhook believes the next steps will be focused on truly achieving the dream of creating that highly coveted 360-degree view of the consumer. By combining demographic information and behavioral targeting, marketers will be able to determine not just which campaign to deliver, but which specific creative materials. Vanderhook illustrates a scenario where marketers can, in real time, identify whether the user is an 18-year-old male who would rather see a Honda with a pretty girl on the ad, or a working mother who would respond better to a Lexus ad with family-oriented features.
Technology is working to ameliorate the age-old marketing dilemma. As Howe states bluntly, "Stop bombarding your clients with crap." When it comes to customer relationship management, he says, "We do ourselves a disservice by basically burning out the consumers because we don't know what they want." In an era supposedly marked by an empowered consumer, marketers are being put to the test: Do it right or don't do it at all. But Howe seems confident that marketing's role in the overall consumer experience is here to stay: "We're going to evolve more, and more money's going to be spent on technology that offers consumers a better indication of a product or service that is aligned with their desires, wants, and needs."
SIDEBAR: How Do You Define Behavioral Targeting?
"We look at advertisers who are looking for a certain behavior undertaken by the user, whether this behavior is a reaction to a certain type of advertisement, using certain sites,
or searching for certain keywords."
--Michael Greene, research associate, JupiterResearch
"Generally speaking, [BT] talks about using behavior to do different kinds of marketing and targeting work, instead of using rules and...other information. Simply by watching user behaviors, you can target and market content, product, or advertising."
--Jack Jia, chief executive officer, Baynote
"Tracking a single user from one Web site to another and gaining info about them, what their interests are in life, and essentially trying to match the ads and
make them relevant to the consumer."
--Tim Vanderhook, chief executive officer, Specific Media
"A lot of the information collected to determine and discern your segment is, in fact, itself behavioral. I don't think there's a perfect definition here, but it's evolving."
--Richard Howe, chief marketing and strategy officer, Acxiom
And a word from an expert about the difference between behavioral
marketing and behavioral
targeting:
"We typically use the term behavioral targeting to describe a tactic that is part of a wider, targeted marketing effort. I personally would try to avoid using the term behavioral marketing because it implies that other marketing tactics (geotargeting, demographic targeting, creative optimization, etc.) are somehow not involved in a campaign."
--Michael Greene
Contact Editorial Assistant Jessica Tsai at jtsai@destinationCRM.com.