The Power of Personalization
Web technologies get personalized to increase profits
- By Clara Parkes
It's a universal phenomenon: The moment you walk into a bookstore, your mind instantly goes blank, erasing any record of what you'd hoped to find there. In the "good old" days, the shop owner would recognize you, greet you by name, and guide you to new releases she knew would interest you. But when you visit a behemoth Web site to make a purchase and forget what you came for, you're on your own.
For years, e-tailers have tried to remedy this problem. Although the solutions are still potentially annoying at times, they're a start. Buy a bluegrass album on Amazon.com, for example, and you'll automatically be shown the latest bluegrass releases the next time you visit the site. Let a friend browse Amazon.com on your computer, however, and who knows what you'll be bombarded with later.
The idea behind such practices is that by better understanding users, online retailers can provide them with a more personally relevant online experience. Beneath this well-intending crusade for customer loyalty is the desire for greater profits.
Personalization is Born
It's called personalization, and it's being incorporated into a range of Web technologies, from Web application servers to content management systems. Web site personalization efforts already demonstrate clear benefits where few other technologies can. According to Jupiter Communications, e-commerce Web sites using personalization technology have seen annual revenue increases of up to 52 percent. Such increases can help justify the average price tag of $50,000 to $3 million for Web personalization systems.
Open-source systems can lower the price curve considerably, as do internally built Java-based systems. But in both cases, these technologies provide only the most basic functionality. Until the more established open-source organizations release their personalization technology (which they're actively building right now) for their Web application and content management systems, the common advice is to stick with over-the-counter technology instead.
Members of the analyst community agree about the value of personalization. Explains the Yankee Group's Lisa Malstead, "The more effort customers put into customizing a site and making it 'their own,' the more stake they have in the site and the higher the chances that they'll keep coming back."
Despite such positive reaction, personalization technology has yet to see widespread industry adoption. According to a recent Yankee Group survey of 200 mid- to large-sized businesses, fewer than 25 percent of those surveyed had personalization capabilities on their Web sites.
Yet those same respondents indicated that the top benefit of an e-commerce initiative is the ability to "effectively target advertisements or advertising messages and offers to customers." Web personalization is key to making this happen.
Current personalization technologies can be classified in three broad categories: User profile-based personalization, rules-based personalization and collaborative filtering. Often the most successful implementations involve a combination of all three approaches.
User Profile Personalization
The least expensive and usually the most accurate approach, user profile-based personalization works when users provide data for their profile, such as name, gender, geographic location and content preference. Such data is most commonly collected during the user registration process, when the "captive" user is much more willing to provide data in order to make a purchase or access a special area of a site.
After registration, however, it's generally difficult to solicit more information from the user. This type of personalization is considered "intrusive" because the user is directly solicited; and "explicit" because the user defines their profile.
In this scenario, organizations establish business rules that define which information is available to which type of user and at what point during their interaction. Rules-based personalization is commonly used within corporate portals. Depending on who the user is, they will have access to varying applications and data.
Likewise, within b-to-b exchanges, vendors can provide different products and pricing schemes to partners depending on the nature of each relationship. From an operating efficiency standpoint, rules-based personalization is the most effective approach.
This is the most complex and expensive personalization approach because more behind-the-scenes technology is required. Collaborative filtering generally entails tracking customer behavior and matching it against the behavioral tendencies of other users (deemed "like-minded" because of preferences, behavior, geographic location, etc.).
Collaborative filtering is most effective for upselling and cross-selling within e-commerce situations. Amazon.com uses it, for example, to recommend other books that might interest users based on their browsing and purchasing behavior.
Collaborative filtering depends greatly on cookies to track user behavior without their knowing it. Because of the non-intrusive nature by which implicit data is collected, however, collaborative filtering is almost never 100 percent accurate. Common categories for filtering include job function, browsing and purchasing behavior, customer loyalty, geographic location and industry.
Is Personalization For Your Business?
IBM identifies these five key indicators of companies that can benefit from Web content personalization strategies:
- Your Web site provides recommended items that are all indexed in a database.
- Your Web developers need to demonstrate an ROI for the site.
- Your site has multiple products (more than 500 product numbers).
- Your site provides e-commerce, customer support or auction functionality.
The scope and functionality of personalization often places it within Web application servers. The major players in both markets already have strong product offerings.
For example, IBM offers WebSphere Personalization, which runs on top of IBM's WebSphere Application Server. It's powered by three main personalization engines: A rules engine, which executes the business rules that determine which content is displayed to whom; a resource engine, which calls upon content and profile information from multiple sources; and a recommendation engine (Macromedia LikeMinds Personalization Server), which supports content and product recommendations via collaborative filtering technology.
Meanwhile, in June 2001, Oracle released Oracle9iAS Personalization as an option to Oracle9i Application Server. The system collects and stores data and builds predictive models within the Oracle9i database before deploying real-time recommendations on Oracle9iAS.
Also in the summer of 2001, BEA Systems augmented its personalization offerings by incorporating analytical and marketing automating technology from Broadbase into the company's WebLogic Personalization Server. Within the BEA environment, WebLogic Personalization Server works alongside BEA's Commerce Server and Campaign Manager to provide almost CRM-style functionality.
You can't implement a Web site personalization strategy without something to personalize: information. For this reason, personalization technology is also popping up within Web content management systems. (To learn more about content management systems, read the September 2001 article, "Get Your Content Under Control!")
Content management leader BroadVision's One-to-One personalization technology is at the core of the Bank of America's newly launched My Bank of America service. The site provides integrated access to a targeted selection of products, content and messages for individual customers.
BroadVision competitor Vignette also offers personalization technology within its Content Suite V6 product.
Personalization and Portals: A Natural Mix
In their most basic state, portals offer single points of entry to vast amounts of information. They were intended to bring order to information chaos by classifying the information into high-level categories and subcategories for easier navigation. Classic portal examples are Yahoo and Excite.
Portals have been one of the earliest and strongest adopters of personalization technology. Many allow users to customize their entry pages to reflect individual interests. The user gets faster access to what he or she wants, and the portal keeps a loyal user. Some portals will even let users customize the underlying look and feel of their pages, including fonts and colors. The idea is that by giving users greater ownership of their information experience, they'll be less likely to pursue other portals.
Retirement planning and benefits information provider BenefitsCorp uses BEA WebLogic Commerce Server and Personalization Server to power its new self-service Web portal. Account executives access the portal from PDAs to obtain immediate, personalized customer information that they use during sales calls.
The portal took less than three months to deploy. Tom Ulrich, director of applications and systems management at BenefitsCorp, explains the process, "We needed an e-business platform that complied with the J2EE standards, could easily integrate with existing systems and demonstrated the scaleability to support a long-term architecture." Ulrich's team decided to use BEA WebLogic Commerce Server partly because of BEA's commitment to Java. But also, says Ulrich, "The personalization features are critical to meet the level of customer service and attention to detail that will drive our business."
Personalization isn't limited to portals, however. It's being adopted in the online retail and media communities as well. In the retail environment, the simplest and most powerful personalization approach is simply to give users the ability to view and track prior orders. Not only does it raise customer satisfaction, but it also frees up customer service staff for more demanding requirements.
Personalization and the ROI Challenge
One of the major challenges today is determining the ROI for Web initiatives. In the previously mentioned Yankee Group survey, 66 percent of respondents said that management doesn't view the Web as important to business objectives because management couldn't measure the impact of the site on the overall business.
The need for accurate measurement methods is very real. Current standards focus on tracking site visitors and increases or decreases in company revenue. Although they still fall short of providing the full picture, they do provide a start.
Take Lehigh Valley Safety Supply (www.safetyshoes.com) as an example. The company has sold its safety shoes online since 1996 but recently overhauled its Web site, adding personalization technology to enable upselling, cross-selling and customized marketing. The system is based on IBM's WebSphere personalization technology. The resulting ability to create customized marketing campaigns has pushed Lehigh's Web sales through the roof, and the company expects to increase online sales by 700 percent in 2001.
Giga estimates that by 2005, more than half of the Internet's users will be non-English speakers. The opportunity for personalization on a global level is thus tremendous—and it presents a whole new set of challenges. According to Giga, "Web content globalization and localization will become a critically important business issue during the next two years as organizations reassess current Web content management practices."
How can companies present tailored Web experiences to an international audience, being mindful of regional languages and cultures while maintaining a globally recognizable corporate image?
Globalization and localization aren't just issues of translating sites into different languages. They require that companies take subtle cultural factors into account, such as specific word choices and the appropriate use of nationally recognized symbols. They even need to choose a color scheme that is respectful of local customs, religions and history.
Just as CDI systems support personalization by maintaining a single customer record, globalization systems provide the international context for personalization campaigns. These systems normally operate in tandem with content management systems, managing the complex workflows that multilingual sites require.
One Customer, One Personalization
Personalization systems are also being situated within the broader context of customer relationship management (CRM) systems. These systems integrate and automate the sales, marketing and customer service processes within an organization. They also provide customer history data, which can be integrated with user profiles to allow for much deeper personalization.
This broadened approach has a few drawbacks. Each new source of customer information must be integrated with existing records to avoid multiple, disconnected instances of each customer. There's nothing more unimpressive than receiving a credit-card solicitation from a company whose credit card you already carry, or having a customer service representative have no record of your online transaction.
Any successful personalization strategy must therefore include resources for building and maintaining a single view of the customer, regardless of where they interact with the business. And the easiest way to maintain this view is by implementing a customer data integration (CDI) system within your CRM environment.
How Personal is Too Personal?
The Web has provided a level playing field in which everyone—regardless of race, religion, gender, financial situation, sexual orientation, nationality, you name it—can enjoy the same experience. Although it may have evolved from good intentions, personalization risks introducing the very same prejudices that drove people from more conventional buying channels to the Web in the first place.
Some personalization practices are so probing that they produce a "Big Brother is watching you" response in users. Instead of entering accurate data during the registration process, some people provide false names and create a completely irrelevant user profile. And here is where even the most costly personalization campaign can prove useless.
Another sometimes problematic use of personalization is to provide special treatment to loyal customers. The concern here is making sure companies still provide a rich enough experience for all other users. If not, such practices can quickly backfire.
First, if companies provide their true value only to prime customers, new users may have no idea it's available. Second, most users won't appreciate being considered lesser customers than their preferred counterparts. And third, if companies don't keep their customer segregation policies above water, they'll quickly face legal challenges. In all cases, they lose potentially profitable customers.
Those in the personalization industry are well aware of these issues. In response, an international advocacy group called The Personalization Consortium (www.personalization.org) was founded in 2000. The group sponsors research, develops industry standards, promotes the benefits of personalized electronic marketing and works with other advocacy groups to ensure the responsible use of personalization technology is in accordance with its proposed Privacy Monitoring and Assurance Program.
Once again, the best defense for individual sites is to think through your personalization approach carefully. Consider each element of data you wish to collect or track, and be prepared to justify it from both a business and legal perspective. If yours is a "what they don't know won't hurt them" approach, be prepared for it to backfire eventually.
B-to-B is the Place to Be
Perhaps the greatest area where personalization technology can thrive without conflict is within the b-to-b sector. Most b-to-b approaches are rules-based, non-intrusive to the user and operate within the framework of explicit—and thus highly accurate—information. The results are greater site usability, higher user productivity and overall greater business efficiencies.
The personalization market will see healthy growth in coming years. According to AMR Research, the CRM segment—including sell-side e-commerce markets such as content management, personalization, channel management and order management—is projected to grow at a healthy 43 percent in 2001 alone. As we move personalization efforts into a globalized corporate environment, the potential for growth is almost limitless.