Stop trying to sign me up! <BR>I'm already your customer!

If a credit company solicits you week after week to become a customer, and you already are one of its customers, how does that make you feel?"American Express badgers me to open an account over and over again, and I've been a customer for 13 years," says Michael Gould, an analyst at the Patricia Seybold Group.
Written by Charles Babcock, Contributor

If a credit company solicits you week after week to become a customer, and you already are one of its customers, how does that make you feel?

"American Express badgers me to open an account over and over again, and I've been a customer for 13 years," says Michael Gould, an analyst at the Patricia Seybold Group. "If they don't know I'm their customer, what do they know about me?"

Gould thinks his American Express experience is about to be repeated for millions of customers online, and not just those of credit-card companies. Electronic retailers, business-to-business marketing companies and electronic bill presentment system vendors are about to try to sell their existing customers products they have already bought - or worse, don't have any interest in buying.

That's because it's hard to compile an electronic sketch of a customer that amounts to the full picture. Even if a profile is drawn, it's still extremely hard to use more than a fraction of it when the customer suddenly appears at a Web site. As of today, there is no electronic shopkeeper at the door, recognizing you as you walk in.

Several vendors, however, are now offering software packages that improve both B2B and business-to-consumer relationships. The SAS Institute, Sagent Technology and Ardent Software - all experienced vendors at mining information, including customer information - are expanding their data mining and data analysis systems and augmenting them with new links to the Web.

The SAS Institute, for example - supplier of the SAS System data analysis software - on Feb. 28 unveiled e-Intelligence, a software combination intended to mesh data about buyers from several sources and make it available to e-commerce applications. E-Intelligence is built atop Enterprise Miner, SAS' core data mining system, but it adds WebHound for clickstream analysis and e-Discovery for meshing data from outside sources into the Web site applications.

E-Discovery, for example, might take credit worthiness information about companies and mesh it with a B2B e-commerce application. No deal above a certain level would be allowed to go through unless a check with a Dun & Bradstreet database gave that company flying colors, says John McIntyre, SAS' director of market strategy.

In addition to credit information, however, e-Discovery can make use of U.S. Census information that characterizes income levels and average size of family based on location, as well as other valuable attributes, once a customer's address has been captured by a Web application. There are also other demographic databases, such as Acxion or Epsilon, that feed information - which is sometimes specific to individuals - to companies that wish to purchase their services.

Data from all these sources, as well as a business' own customer relationship management system or financial systems, can be loaded into the data warehouse, searched and analyzed by the Enterprise Miner/e-Discovery systems and presented in summary form by e-Discovery to the Web system, McIntyre says.

"It does most of the work offline [before the customer comes to the Web site]. Then when the customer shows up, the Web application can say, 'Here are the things that we think best fit your preferences,' " he says.

Discerning customer preferences is the Holy Grail of e-commerce, given its impersonal and error-prone conduct to date. In some cases, customer preference is decided by an instant analysis of where the customer goes on the site. Either that clickstream information a-lone, or a match against previous visitors who followed that path, is used to determine a customer's likely preference in goods and services.

"It is not yet known how reliable clickstream data is for predictive purposes," McIntyre says. That's why SAS is concentrating on bringing other data sources into the picture, and doing so in a way that is presummarized for quick feeding into interactive Web applications. Achieving that combination of circumstance - information that is both in-depth and quickly available - is neither easy nor inexpensive: E-Discovery's base price is $200,000, he says.

Another option for quick information on the customer is Sagent's Address Cleanser & Coder Transform, a product name that Michael Ault, director of corporate communications, apologizes for, given its lack of imagination. He says the product, announced Feb. 24, was inherited in Sagent's acquisition of Qualitative Marketing Software last year. But it describes what it does, he adds.

Almost 40 percent of the addresses entered in online purchases prove to be problematic - lacking in "Avenue" or "Street" or "Drive" after the street name in cities where there is more than one street with a given name, or having some other problem. To reduce the volume of returned goods and dissatisfied customers, Sagent's Address Cleanser & Coder Transform takes the customers' addresses and checks them against the U.S. Postal Service database. It corrects addresses and fills in missing components when customer name and location can be verified, says Ellen Raynor, senior product manager at Sagent.

At the same time, Address Cleanser & Coder Transform loads the customer database with information on that address from U.S. Census and other sources, often giving an economic profile based on the income level of the neighborhood and other information about the customer.

Address Cleanser is available immediately as a software package and priced at $50,000 to $100,000, depending on the databases ordered.

Ardent, which is in the process of being acquired for $880 million by Informix, announced enhancements to its DataStage Suite on Feb. 23. DataStage is a Web data analysis system and data integration package.

DataStage now includes ClickPack, which allows the DataStage to extract data from Web server log files and e-mail systems and parse it for user information. ClickPack can also be extended by customers using Perl scripts, says Mikael Wipperfeld, vice president of data warehouse marketing at Ardent. And DataStage includes XML Pack, which allows DataStage to read the eXtenisble Markup Language, create XML documents, and import and export XML metadata that summarizes business information and documents.

DataStage also includes a plug-in for IBM's middleware MQSeries, which can integrate a Web application with numerous back-end functions, such as database applications, mainframe file systems or Enterprise Resource Planning applications, Wipperfeld says. The core DataStage software is priced at $140,000, with ClickPack, XML Pack or other add-ons priced at $15,000 each.

Seybold Group's Gould says attempts by Ardent and other vendors to align the internal business systems with the outward-facing Web applications, and coordinate the customer information in each, are headed in the right direction.

So far, Gould says, most businesses have been preoccupied with capturing the superficial information available in a customer's first visit, and the deeper sets available through previous transaction, service calls and so forth has been harder to coordinate and integrate. Even though strides are being made, Gould says, "everybody still has a long ways to go."

Editorial standards