Custom Fit

Personalization can greatly improve productivity and usability while providing key marketing advantages

By Colin White

Easier-to-find information. Better e-business application usability. Improved productivity. More focused working client relationships. These are the benefits of personalization for consumer and business users of the Internet and intranets. In today's competitive market you cannot afford to ignore personalization applications. These applications customize the Web content consumers and business users view through their desktop and mobile online interfaces.

In this article, I will explore personalization from two perspectives - the consumer and the business user. I will first examine how consumer e-business applications employ personalization to customize Web storefronts to create a one-to-one e-marketing experience for Internet users. Then I will discuss the role of personalization in the corporate environment for providing front-office users and trading partners with a personalized view of business content that enables them to do their jobs and run business operations more effectively.

Personalization for the Consumer

In the consumer e-business marketplace, personalization can be driven by Internet users themselves or by the e-business applications that interact with users. An example of user-driven personalization is my.yahoo.com where you can create a personal profile of the information resources (news feeds, weather, stock prices, and so on) that you wish to see displayed on the home page when you first connect to the Yahoo site. Users do not have to navigate the generic Yahoo home page to find the content they require - instead, the Web server tailors the displayed content based on each user's profile. For online shopping, Yahoo also has a My Shopping profile that enables consumers to manage orders and to specify the online retail stores they like to visit.

User-driven personalization is typically used for filtering content such as news feeds, rather than for the one-to-one e-marketing of products and services. The main reason that it is not used for marketing products is that consumers are rarely motivated to create profiles detailing the types of goods they like to purchase. Nonetheless, there are some exceptions. Lands' End, for example, offers My Virtual Model, in which you enter your measurements and then try clothes on a 3D model. Over 1.5 million people have entered their measurements since this option became available on the Lands' End Web site. In general, however, if you are selling products over the Internet, your e-business applications must either learn about the user's preferences over a period of time, or place each customer in a particular market segment, and then use marketing data for that segment to personalize Web pages of product offerings.

Inference-Based Software

E-business applications employ either inference-based or rules-based techniques to personalize the products and services sold over the Internet. Inference-based software is usually used for cross-selling when a business has a wide range of different products and product lines to market on the Web. It is also useful when little is known about customers and their preferences. With an inference-based approach, the software tracks, or profiles, a Web user's behavior, identifies other people (sometimes called mentors) who have a similar behavior, and then uses these mentors in conjunction with a recommendation engine to create product suggestions that may be of interest to the current Web visitor. Amazon.com is a good example of this approach.

After purchasing a product on Amazon, the Web user is shown a list of other products that were bought by people who purchased the same product. One problem with the referrals, of course, is that a product cannot be recommended until someone else has already bought it. Products that provide inference-based recommendation engines include Macromedia LikeMinds and the Recommendation Engine in Net Perceptions' Personalization Manager. These products are often included with Web application servers, for example, IBM WebSphere (LikeMinds) and Vignette V/5 (Personalization Manager). For a brief discussion of overcoming opposition to personalization, see the sidebar "Privacy Matters" in the online version of this article on IntelligentEnterprise.com

Rules-Based Software

In their simplest form, rules-based products are used during marketing campaigns to sell specific products or services. When the user moves to a particular Web page, the latest special offers associated with that section of the Web site (consumer electronics, for example) are displayed. The information displayed is governed by rules that are defined by marketing staff based on business expertise, Web-site analysis, and experience gained from earlier campaigns. Web application servers from vendors such as BroadVision Inc. (Retail Commerce Suite), IBM (WebSphere), Microsoft (Site Server), Open Market Inc. (EBusiness Suite), and BEA Systems (WebLogic Server) all support a rules-based approach to building dynamic Web pages. Several vendors provide Web clickstream analytic applications that help marketers to understand buying patterns and to fine-tune their marketing campaigns. Vendors with products in this latter category include Accrue Software Inc., Macromedia Inc., Broadbase Software Inc., E.piphany Inc., Informix, Informatica Corp., Net Perceptions Inc., NetGenesis Corp., and WebTrends Corp.

Focus on the Customer

The objective of e-marketing is to create and retain loyal customers that become regular visitors to an online merchant. To achieve this loyalty, personalization must be customer-centric, not product-centric as is the case with many Web sites. Thus personalization must focus not only on providing tailored product offerings, but also on providing better service. One approach to customer-centric personalization is to capture and record every online purchase made by the Web user. Each record can then be analyzed and used to create a detailed profile of every customer's buying preferences. The more in-depth your profile is, the better your e-business application can personalize your Web content to meet customer needs and provide better service. Accurate data about customer preferences enables e-business applications to use techniques such as rules-driven alerts that inform customers via email about new product offers or recalls, or in the case of investment firms, about events that may affect the value of stock holdings.

Nevertheless, a true customer-centric approach to marketing does not stop here. A business should track every interaction with the customer, including calls to service and support centers and purchases made via other channels (by telephone or in a retail store, for example). This information is maintained in an integrated customer database (sometimes called a customer operational data store, or customer ODS). The challenge in creating an integrated customer ODS is correlating interactions that may have different customer identifiers at each touch point.

An integrated customer ODS can give you a detailed understanding of your customers - enabling you to market and sell through sales channels your customers prefer. Your business can improve customer service by identifying after-sale issues such as delayed deliveries, product problems, and so forth. Retaining customers involves not only having a personalized interface, but also providing superior personal service. Data mining and OLAP technologies can be used to analyze the contents of the ODS to determine customer buying habits and trends. This analysis can then be used to drive campaign management systems.

It is not possible for a single vendor to develop a complete set of products that support e-business operations, analytic applications, a customer ODS, recommendation engines, and automated rules generation. Instead, vendors are forming product development and marketing relationships that support the intelligent e-business framework shown in Figure 1.

Personalizing Corporate Portals

The fastest growing personalization technology in the corporate business environment is in the area of corporate portals. The main objective of a portal is to provide business users with an integrated and personalized view of corporate business content (information, applications, expertise, and so on). Portals are a hot topic; and vendors have been quick to jump on the portal bandwagon with the result that the marketplace is crowded with portal products at different levels of maturity.

Portals have their origins in the Internet where search facilities like my.

yahoo.com provide users with easy access to information. Corporations have rapidly adopted portal technology for organizing and personalizing access to information on corporate intranets because it dramatically improves user productivity. In many organizations, a portal has now become the sole interface to information managed by both internal and external Web servers. Over the past year, portal products have evolved beyond simple interfaces to Web information and now support additional business content like business intelligence, workgroup information (documents, email, and so forth), and back- and front-office applications. Examples of representative portal products include IBM Enterprise Information Portal, Informix's Axielle E-Intelligence Portal, Lotus K-Station, Oracle9i AS Portal, Plumtree Corporate Portal, Verity Portal One, and Viador E-Portal.

To date, most portal products have been employed by internal business users, but with the growth of e-business use, organizations are now integrating portal technology into their e-business systems. This integration allows a company to extend portal access to external trading partners and key clients, which improves business relationships, and helps optimize business processes such as supply chain management. Integrating portal technology into the e-business environment is a complex task. It requires a portal that not only supports internal e-business applications, but also interoperates with external systems. Extensible markup language (XML) is beginning to play a key role in helping solve the problem of application integration and information interchange.

Portal products use a variety of different approaches to personalize the content that business users view and access via a portal's Web interface. As in the consumer environment, these approaches fall into one of two broad categories: user-driven personalization and application-driven personalization.

The User-Driven Approach

With this personalization approach, portal users are presented with a list of business content sources, and they select what that they would like to see displayed on their personalized portal home page. This is somewhat similar to the personalization facility supported by my.yahoo.com that was discussed earlier. The list of sources is created and maintained by portal administrators and can be updated by users publishing business content to the portal for sharing with other portal users. An entry in the list may be the name of an application or document, or may be the name of a business content server (a database, file, or document management system, for example) that manages content on behalf of a specific application or department. To help locate information managed by a business content server, a portal will normally provide a search facility.

The main issue with the user-driven approach to personalization is that there is usually little ability to categorize source content by business subject area - other than a rudimentary facility that organizes the list of content sources by keyword. The lack of a rich categorization facility makes it difficult for portal users to locate needed content as the number of content sources increases. To solve this problem, some portal products provide a categorization manager to document information about source business content in an information directory. Entries in the information directory are organized in a hierarchy of subject-based folders where each subject is related to a specific business topic or concept. The business taxonomy that defines this folder structure is developed during the portal design process and is used by the categorization manager to determine the most appropriate folder in which to document information business content.

Categorization is done by applying taxonomy rules to information extracted from the business content itself, or from the metadata about the business content (file name, Web page URL, author, and so on). Sometimes external search engines are used to help create or enhance the business taxonomy. Examples of search-engine vendors include Autonomy Corp., Semio Corp., and Verity Inc. Products from these vendors use a variety of techniques (pattern matching, rules-based, and semantic analysis) to search source content and to create a business taxonomy.

The Application Approach

A portal's categorization manager and its associated information directory support an application-driven approach to personalization. Here the information directory becomes the roadmap to the complete domain of business content that can be viewed by business users through the portal. To find the business content they require, portal users navigate the folder entries in the information director, or search the directory using the portal's search engine. Information directory entries seen by the user are filtered based on security requirements, and on personalization rules that define the type of business content that users require to perform their jobs. Not all portal products provide a categorization manager or an information directory, and this is a key distinguishing factor between portal products.

As this brief article has shown, there are many different uses for personalization and correspondingly many types of personalization technology and products. Used wisely, personalization can provide significant productivity gains and can become a key marketing advantage for e-businesses.

Colin White (cwhite@databaseassociates.com) is the founder of DataBase Associates and specializes in data warehousing, business intelligence, database systems, and intelligent e-business. He is a frequent contributing writer to Intelligent Enterprise and other magazines.