Booking e-commerce lessons

IT IS NOW well over five years since the Internet made its first inroads into the homes of American consumers, and e-commerce was born. Since then, many dot-coms have come and gone, as has the early volatility of the market. Today's e-commerce players feature a more measured range of products and services from companies with recognizable brands. Consequently, for established e-commerce businesses, the challenge today is not so much about staying viable, but more about staying ahead of the competition by providing a fresh, compelling experience for the customer.

Free IT resource Hear how top CIOs turn change into a competitive advantage. Sponsored by HP Free IT resource Try Sun servers, workstations and storage products free for 60-days. Sponsored by Sun Microsystems

For the CTO, this represents a critical challenge. Although there is no shortage of opportunity for adding new Web site features or the supporting technologies, we must set a strategic course that promotes business growth without allowing our systems architecture to grow out of control. To remain manageable and cost-effective, systems architecture must retain a reasonable level of structure and cohesiveness. This takes discipline, because there are countless "solutions" addressing such well-known "problems" as content management, CRM, and Web site analytics.

In each case, we begin with a real business need, are seduced by industry standards, wooed by vendors bearing "solutions," engage the inevitable consultants to provide the necessary customization, and wake up one morning with another strange-looking limb grafted onto our systems architecture diagrams. There must be a better way.

I believe the answer lies in a single-minded focus on the original business goal. Let's examine the problems mentioned above. Content management, for example, is often less expensive and more effective when home grown. At TheStreet.com where I worked prior to coming to Barnes&Noble.com, we demonstrated that an in-house custom solution provided better results at substantially lower cost than the original system built using a well-known content management product.

In July 2002 after four months of development, we deployed our third-generation content management system, developed completely in-house and closely integrated with our core business processes. In each case the application was designed from the ground up based upon the needs and requirements of the in-house creators of content. Although we are not yet using them, our recent work is allowing us to explore Web services.

CRM is another well-intentioned application wreaking havoc with systems architecture. These systems invariably require a database implementation that imposes the CRM vendor's view of the enterprise upon us. From a seemingly reasonable beginning, the typical CRM implementation project becomes a nightmare of custom data transforms, ancillary data feed processes, and missed deadlines. I recommend a clear-eyed assessment of the real needs of the business. What do you want to learn about your customer? In many cases you will find that the data you need lies ready-to-use in existing databases.

Web site analytics is an interesting area in which technology capabilities currently far exceed the needs of even the largest e-commerce Web sites. The danger here is a loss of focus. Sure, I can tag every one of my pages so they report into a central server when loaded, and I can license a service that will create rich graphics that portray site activity in multidimensional splendor. Or I can choose to use a local data collection solution, which will require that I invest heavily in hardware to load and process the massive flood of site activity data. With so much raw data, it is easy to lose sight of the business goal: understanding in a timely fashion how customers are using the Web site or determining the effectiveness of a marketing promo.

My recommendation: Start with a clear and narrowly focused definition of what you need to measure. In many cases you will find that everything you need is already in your Web server log files. Any good scripting language can be used to extract the desired information. While you are at it, collect some statistics about Web server behavior and feed that back to your systems administrators, to help them tune the Web farm. You won't be spending a lot of money, everyone will benefit, and you can maintain consistency with the rest of your technical architecture.