In-Depth

Case in Point: Predictive Web Analytics and ROI

The Financial Times uses Web analytics tool to better understand its customers

• By Stephen Swoyer
• 06/04/2003

FT.com receives 55 million page views per month, with more than 3 million unique visitors—not all of whom are subscribers.

The only problem, confirms Chris Catchpole, head of database services with FT.com, was that his company knew very little about its site visitors. At a time when FT.com was marketing a new subscription-based service, Catchpole allows, this was something of a liability. “One of the major things that we needed to do was [develop] propensity scores for individuals based on the usage of our Web site. The frequency of what they look at, trying to understand their propensity to purchase, to renew,” Catchpole explains.

To do that, Catchpole says, FT.com needed to augment its existing tool, a Web analytics solution called NetGenesis. Although NetGenesis could perform some analysis of customer browsing habits, it lacked the predictive modeling capabilities. The result, Catchpole concedes, was that FT.com wasn’t getting the ROI that it had hoped from the NetGenesis software. “We were having real difficulties in terms of getting down to the granular level, as well as being able to use this for marketing or site design or any of these areas."

The good news for FT.com was that predictive analytics powerhouse SPSS Inc. acquired NetGenesis Corp. in October of 2001. With a 30-year history in statistical and predictive analysis, along with extensive experience serving the retail, financial services, and telecommunications industries, SPSS brought credibility to the burgeoning field of Web analytics.

Recently, FT.com implemented Predictive Web Analytics, a new analytic and data mining solution from SPSS that combines the NetGenesis product with SPSS’ Clementine predictive modeling and data mining software.

SPSS has struggled to recast its often-sophisticated analytic products as tools that are suitable for use by mainstream customers in non-traditional industries. Its new series of Predictive Analytics applications, of which Predictive Web Analytics is the second deliverable, is the Chicago-based vendor’s most ambitious such attempt to date.

Jay Henderson, senior product manager for Web analytics with SPSS, says his company’s Predictive Analytics applications exploit its traditional strengths in statistics and data mining, largely for desktop users. “One of the trends we’ve seen in the web analytics market space is that as this space matures, the types of analysis people want to do have matured. That’s where we’re pretty well positioned to bring this solution to market."

The result is a solution—Predictive Web Analytics—with a compelling ROI. “This type of analysis typically has very high ROI, mostly because these projects done with the Predictive Web Analytics solutions typically have well-defined business goals, such as understanding which groups of customers are your most profitable ones, understanding different cross-sell opportunities, things like that,” Henderson asserts.

For FT.com, an existing NetGenesis customer, the ROI angle was indeed tempting—but SPSS dangled another, almost irresistible, carrot: “What we’ve done is set up a deal with SPSS where we are trialing the Clementine software to show the business value that we can get from this kind of statistical tool,” Catchpole says. Moreover, because Predictive Web Analytics is enabled by means of components that facilitate integration between NetGenesis and Clementine, Catchpole says that FT.com didn’t have to make drastic changes to its environment. In this respect, and in view of the ROI potential of Web analytics, SPSS’ offer was a no-brainer.

What kind of business value does FT.com expect to realize from its use of Predictive Web Analytics? According to Catchpole, his company’s Web sites offer two varieties of services: Free content that’s available to all users and premium content that’s available only to subscribers. Many of FT.com’s first-year subscribers will soon be given the option to renew their subscriptions, and the company wants to better understand how the usage patterns of its readers affect their renewal rates.

“We’re looking very closely at the people who are due to renew in the next few months—what are their usage patterns. We’re trying to watch the first month of our renewals come through and calculate on the back-end how the usage patterns have affected the renewal rate. Are they less or more likely renew depending on which sections they use?” Catchpole explains.

FT.com also wants to be able to better tailor its content, messaging, and advertising to suit the interests of these readers. Catchpole offers an example of a reader whose usage pattern suggests that she’s primarily interested in IT-related news: “What we would be able to do is not necessarily target them with IT advertising when they visit that area of the site, but also wherever they are on any area of the site, they’ll see IT advertisements.”

Then there’s the issue of readers who haven’t yet subscribed to FT.com’s premium service. For obvious reasons, Catchpole allows, FT.com is very interested in these readers, and wants to understand what characteristics, if any, are shared among visitors who choose to subscribe to premium content. “We have a number of people who still use our Web site who don’t subscribe, and we’re trying to work out a propensity score for the current subscribers. Through the use of this we hope to be able to predict with more certainty a number of ways to get more commercial value by enticing more subscribers,” he suggests.

FT.com is currently in the midst of its Predictive Web Analytics implementation, and has already started to integrate its NetGenesis system with Clementine. In this respect, Catchpole confirms that FT.com’s implementation has thus far been “fairly pain-free,” but allows that his company has undoubtedly benefited from the fact that it already has NetGenesis in-house.

At the same time, Catchpole points out, customers using a Web analytics solution other than NetGenesis can also exploit SPSS’ predictive modeling capabilities. “You can leave the structure of your Web analytics database behind with the possibility of using SQL scripts to extract data and feed it into Clementine,” he suggests. “I don’t think that [NetGenesis and Clementine] necessarily have to be used together, but I think that you’ll see quicker ROI.”

SPSS’ Predictive Web Analytics is available for $135,000, and that’s just for starters. Catchpole says that he believes that the ROI is there. “My suggestion is to take one or two avenues that you can quickly prove ROI and demonstrate the commercial impact that such products can have on your marketing and on your overall bottom line. In that way, it’s easier to get backing for the investment internally to build up going forward.”

In this case, he concludes, FT.com expects to start realizing ROI on its SPSS investment in the coming quarter, as it more effectively targets its online subscribers for renewal. “With our renewal mailing we would expect to see our renewal rates increasing slightly through the use of being much more targeted in our marketing based on the better use of the site,” he asserts. “That’s only the first part of it. As we begin to deliver targeted content for our subscribers, and as we entice new subscribers based on what we’ve learned, the ROI will be there.”