NEW YORK -- Data warehouses -- those massive repositories of information that retailers draw upon to improve everything from customer relationships to supply-chain processes -- can become even more functional when stocked with information from on-line interactions.
That's the thinking of Ralph Kimball, president of Kimball Associates, Boulder Creek, Calif., who said the Internet holds great potential for companies seeking to leverage their substantial investments in data-warehouse technology, while at the same time greatly enhancing their customer-relationship management abilities.
An expert in data-warehousing technology, Kimball presented the keynote session, "The Marriage of Data Warehousing and Customer Relationship Management," during the Customer Relationship Management conference and exposition held here earlier this month and sponsored by DCI, Andover, Mass.
Recognizing the value of analyzing data drawn from all aspects of their operations, retailers are increasingly upgrading the capacity of their data warehouses. Wal-Mart Stores, for example, owner of the world's largest retail data warehouse, is expanding its size from 44 terabytes to an unprecedented 101 terabytes, allowing the retailer to store as much as two years' worth of sales data.
What some retailers may not yet see, however, is the great value that data gleaned from the Web can bring to their data-warehouse projects.
"We need to grab this big data resource that has landed on our front step: the click stream," said Kimball. The "click stream" - the navigational path on-line consumers follow as they wend their way through Web sites -- contains data that can reveal new insights about customer behavior.
Retailers that collect click-stream data, feed it into their data warehouse and analyze it, stand to learn more about their customers than was ever possible in the physical world.
"The click stream is more than [a record of every on-line] transaction; it is every gesture leading up to a transaction. It is as if we could see the customer 'walking' up the aisle, and there may be a hundred gestures for every purchase," Kimball said.
Kimball said traditional retailers who are tuned into customer service, such as Nordstrom, are very aware of Web-user tracking, but haven't made it the dominant theme in their business. "After all, they have billions tied up in pretty successful real estate. So their strategy may be a hybrid -- to draw people physically to their stores," Kimball told SN following his presentation.
There is no limit to the number of ways companies can segment and characterize their customers using click-stream data, he said during the presentation. For example, conventional demographic data such as age, household income and education can be enhanced with behavioral attributes such as how an individual enters a Web site -- whether it's through a search engine or another Web site -- or what types of interactive technologies engage, or alienate, customers.
"All those things are measurable," he said.
Kimball said some companies are assigning hundreds of attributes to characterize a single customer; others have even gone so far as to assign 1,000 attributes to a customer.
Armed with more detailed information about their customers, companies can then promote more effectively and in a highly customized manner, enhance products and services, and learn how to improve their Web site design to be more intuitive and user-friendly.
"It's clear that customer-relationship management is being defined by the Web," Kimball said. "The Web is sending us many messages about what works and what constitutes good practice. Data warehouses are rapidly becoming 'Web warehouses."'