News

SPSS asks: What will the Web do next?

• By Lana Gates
• May 21, 2003

With Predictive Web Analytics, Chicago-based SPSS has moved to better integrate the NetGenesis 6.0 Web analytics platform with the predictive modeling capabilities of Clementine 7.0. The goal is to provide a way for companies to better understand their customers' Web behavior and to predict how those customers will behave in the future.

Predictive Web Analytics uses data mining algorithms such as clustering techniques to automatically find patterns in data based on behavior. NetGenesis loads Web log data and then goes through it and enriches it.

"The data you need for normal Web analytics is insufficient for data mining," explained Colin Shearer, SPSS's vice president of customer analytics. "Data mining needs richer data."

The London-based information Web site FT.com, an Internet partner of the Financial Times newspaper, is using Predictive Web Analytics to take subsets of Web data and run them through Clementine to find important trends. The site generates 55 million monthly page views through its 3.5 million unique monthly users. FT.com uses trend data as a basis for market segments and product targeting, according to Chris Catchpole, head of database services.

Catchpole further explained that Predictive Web Analytics analyzes a group of users by looking at their Web site behavior over the past three months before their renewal date, for instance. It then takes that information and clusters it into segments. FT.com sends a certain message to those people who are using the site just a little to let them know what they are missing. People who are using the Web site all the time, but that are not using its full functionality, are sent a different message. Those users taking advantage of the Web site's full value receive a still different message.

"We're trying to understand the trends of our usage and how parts of our site are connected and whether people use A and B together or how they navigate around the site so we can understand what they're taking advantage of and what they're missing," Catchpole noted. "It gives us an opportunity to deliver what people need."

Predictive Web Analytics "adds understanding and takes that forward. It opens up new ROI opportunities," SPSS's Shearer said, by allowing companies to predict which customers will buy. "It opens up an avenue for finding online and offline data sources," he continued, giving companies a "more holistic view" of their customers.

Pricing starts at $135,000 plus services. Predictive Web Analytics supports Windows 2000, Microsoft SQL Server 2000, Sun Solaris 7 and 8, and Oracle8i Standard or Enterprise editions.