ZIFFPAGE TITLEHow It Was MeantBy Edward Cone
Is Web Persona-fying Better Than Personalization?
Andrew Peterson knows pretty much everything there is to know about a select group of customers at Sovereign Bank. "We have their age, location and education level," he says. "We know their personality types." In fact, Peterson, vice president for the Internet and emerging technology at Sovereign, a Philadelphia-based financial services company with $59 billion in assets, even has access to photos and psychographic profiles on each one of the customers.
Peterson knows, for instance, about the stresses on Anna C., a single mom with too much to do and not enough time to do it in. He's up on the latest news from Robert and Holly, a young married couple who are working hard to build a stable financial future. And he keeps tabs on Yin, a fun-loving student at Brown University, and on Richard and Debra, who have kids in college and are looking for ways to maintain their affluent lifestyle into retirement. "We know their relationships with Sovereign and with the Internet, what they want from the site and how we can help them reach their goals," says Peterson. "And we use that knowledge to tailor the functionality and content of our site."
Sovereign, it seems, has realized the long-promised goal of personalized marketing, establishing one-to-one relationships with its customers over the Web. Except it's all unreal. None of these customers exist. They are characters, called personas (the marketing trade's term of art for such imaginary people), created by Sovereign and marketing firm Agency.com Ltd., in order to make the bank's Web sites more accessible and productive for actual peopleand more profitable for the company.
Each persona is a stand-in for some segment of Sovereign's customer base, a composite drawn from hours of research and carefully crafted to create what Peterson calls "an empathetic view of the customer." Products likely to appeal to an established family like Richard and Debra's, for example, are pitched and positioned differently than the starter accounts that are more suitable for Yin, the college student. "Decisions on imagery, fonts, language and navigation are geared to those particular individuals," says Peterson.
These personas are key tools in a sophisticated kind of market segmentation that makes use of the Web's depth and breadth to refine the customer experience. But the approach is a far cry from the kind of personalization touted just a few years ago by software vendors and consultantsand pursued by companies to the tune of hundreds of millions of dollars. The next big thing in personalization is less, well, personal than yesterday's overhyped one-to-one methods. But this time it seems to work.
ZIFFPAGE TITLEHow It Was Meant
How It Was Meant To Be
In the beginning, there was Amazon.com Inc., and it was really good at crunching customer data to sell more products. "They have been at it for a long time, accumulating a lot of data," says Harley Manning, an analyst at Forrester Research Inc. "They don't talk about what they do, because that's their competitive advantage." But personalization's second act has been a long time coming. "Ten years later, every conversation about personalization still comes back to Amazon," says Rob Cosinuke, president of Digitas Boston, an office of the national marketing firm Digitas LLC. "What does that tell you about the state of the market?"
Amazon, with its huge investment in technology and an enormous customer database to mine, turns out to be a model that few companies can successfully emulate. "That kind of personalization is the highest fruit on the tree," says Matthew Berk, a former research director at Jupiter Research whose 2003 report, "Beyond the Personalization Myth: Cost Effective Alternatives to Influence Intent," helped quell the personalization hype.
Hype had been in plentiful supply. "There was a giddy, Internet bubble interest in personalization," says Manning. For a while it was a feature everyone had to have." But that vision of personalized marketing, first popularized in the 1993 book, The One to One Future: Building Relationships One Customer at a Time, by Don Peppers and Martha Rogersand pushed by software companies such as BroadVision Inc., Epiphany Inc. and Blue Martini Software Inc.was not for everyone. "It was a beautiful conceptual framework, but you can't go to the run-of-the-mill business and make it work," says Berk. "When this started, even companies that were able to do personalization to some degree didn't know how to measure their results. Nobody really knew what was going on. We didn't understand online behavior."
Berk, who went on to found Open List Inc., which tracks customer preferences by doing targeted searches for the restaurant and hospitality industry, remains skeptical of one-to-one personalization for most companies. "Unless you are Amazon, it's a silly thing to fantasize about," he says. "It is incredibly valuable, but only at scale. For 90 percent of all other folks doing business on the Web, it is a fantasy." There are better ways to spend your money, he says. "When you line up all of the possible investments to make in order to help a business do better on the Web, personalization is last on the list. You don't need the complicated things."
The result: Setting out to really know your individual customers, and then guiding them to buy more stuff by building their personal preferences into the online shopping experience, has gone out of style. Says Manning, "We still ain't there."
Recently, Forrester has been seeing an upsurge in client interest in some less ambitious but more productive methods of personalization. Says Rhiannon Rail, director of participant marketing for John Hancock Retirement Plan Services, which used personas to redesign its retirement benefit Web site: "Initially, we looked at Amazon and said, 'Hey, that's where we want to get to.' But when we realized the volume of data needed, we decided that we didn't want to get into that kind of data-management business. You can get there if you get the data, but the cost can be prohibitive."
ZIFFPAGE TITLEUnreal Customers
Personasnobody seems to use the Latin plural, personaeare one way of creating what Rail oxymoronically describes as "generic personalization." John Hancock, which last year merged with Canadian financial services firm Manulife Financial Corp., built its retirement planning Web site around three personas. "What we are doing is marketing to personalities," she says. "When you talk about 401Ks and retirement, that is an emotional subject. A site needs to be somewhat personalized to be relevant and get customers engaged."
To prepare its personas, John Hancock worked with specialized researchers, including a demographer and the marketing firm Razorfish (now Avenue A Razorfish). The researchers talked to actual customers about what retirement planning meant to them. They found that people in differing financial situations often approach the planning process in different ways, but tailoring the Web experience to a user's wealth was not the way to go. Instead, the site is designed to meet the emotional needs of three different personality types, personified by characters named Rozi, David and Ernest. "Rozi is looking for guidance, but she wants to arrive at her own conclusions," Rail says. David, on the other hand, wants the answers spoon-fed to him, and Ernest (who enjoys cigars, watches The Sopranos, and thinks his siblings resent his financial success) likes to dig for his own information. "We developed the site with three levels of interaction to meet the learning styles and comfort levels of individuals," says Rail.
All customers see the same basic content when they come to the John Hancock site, but their experiences may diverge as they click for more details. In effect, users personalize their own sessions. This kind of transaction-based personalization, in which the conduct of a customer during a particular interaction with the company influences the remainder of that session, is another key element of current personalization strategy. Readers of the Wall Street Journal's online edition, for example, might be shown particular ads based on the sites from which they clicked to the newspaper. A mortgage company ad, for instance, might appear for viewers who have linked from a real estate or lending site.
"People are using personalization to make each session better, based not on what you bought last month but what you are looking at now," says Rich LaFauci, a senior vice president at Digitas New York. This technique doesn't require sophisticated tools such as collaborative filtering to mine clues about customer interests from the context of a session, and then guide customer behavior and experience accordingly. "Session context personalization has a good effort-to-benefit ratio," says LaFauci. "It requires no effort from the consumer, and a much smaller investment in technology than a big database."
At Sovereign, the focus is on increasing the amount of self-service by each customer during a visit to the site. That might mean transferring money, changing an address, or switching from one account type to another. Like personas, self-service helps the bank anticipate customer needs, says Marianne Doran-Collins, senior vice president and director of online and affinity banking at Sovereign. Data gleaned from these transactions gives the company more information about real-life customer behavior. "It lets us build out our view of what they are likely to do online," she says.
Sounds neat, right? So did the first round of personalization. Does any of this stuff actually work? Rail says there is evidence that suggests the John Hancock site has become more effective since its persona-fied redesign was launched in 2003. John Hancock customers use the site much more frequently than the industry average, according to survey data provided by the company, and 84 percent of those who come to the site try out John Hancock's Internet planning tools (versus an industry average of 49 percent). Repeat visits to the site more than doubled after the redesign.
All the while, John Hancock is accumulating information on actual customers who visit the site. Marketing to individual customers remains an aspiration for many companies, and incremental strategies, such as personas and self-tailored sessions, are also ways of moving toward that goal. "Most of the time, especially in the first meeting online, or through a toll-free number, you don't have the data you need to make it personally relevant," Rail says. "We tell our customers that the more we know, the more we can improve their experience. People are more willing to cooperate when you set expectations up front."
Says Digitas's Cosinuke: "Before you can create an individual experience for each customer, you need to make it worthwhile for the customers to share their information with you. Most companies are not doing that."
True personalization is still far off for most companiesand not everyone thinks it is even worth pursuing.
ZIFFPAGE TITLEWhy Get Personal
Why Get Personal?
One problem with personalization is that it is extremely complex to do well. Even basic technologies such as databases have to be scaled to enormous proportions, and the actual crunching of data to yield useful results gets complicated in a hurry. "It is trivially easy on the tech side to let people do things that are absolutely fatal," says Manning. "Creating lots of rules for dynamic content delivery and management is a nightmare. Four rules are more than twice as complicated to manage as two rules, and we saw projects with hundreds of rules. You have to set up some pretty complex sets of algorithms to manage that, some serious programming intelligence, and that is well beyond what was built into personalization systems."
Many customers aren't interested in playing along. "With one-to-one personalization, it is very difficult to service customers, and very easy to aggravate them," says Sovereign's Doran-Collins. A lot of people resent and resist corporate efforts to learn more about them. It's a common problem for Web marketers. "We do large-scale surveys of Internet users, and they say time and again that they lie online," says Manning. "Reason No. 1: They don't want to be marketed to. That's why databases are full of users named Donald Duck and Mr. Fantastic and Mickey Mouse."
Privacy and trust issues are limiting factors on personalization with John Hancock customers, too, says Rail. "I don't need to know everything about you, and I don't need a huge investment to tell me a lot of the things I do want to know," she adds. Companies that interact repeatedly with customers over time can begin to personalize offerings for them in a mutually beneficial way, says LaFauci. "Supermarket couponing, where they print a coupon specific to you based on a set of transactions, is effective," he says. "But most companies are not in a position to have access to that kind of data to be effective, much less cost-effective."
www-marketing.wharton.upenn.edu The Wharton School's online repository of marketing research, including current papers
Frequently updated Web site about marketing tools and strategy, founded by former Stanford professor Allen Weiss
Peppers, coauthor of the seminal book on personalization and a consultant on customer relationship management, says the death of one-to-one marketing should not be exaggerated. "There was artificiality and market hype, and then rationalization in the market," he says. "But we are on a course of no return in terms of increasing personalization. It's not for every product or businessyet look at the way banks personalize services for their most valuable customers. The question is, how to routinize some of that to make it standard for all customers."
But Manning is more skeptical. "The premise of personalization is that the system is smart enough to know what this person wants, to anticipate her needs and push things to her without action on her part. That's a fallacy. The system is not going to be that smart."