BACKGROUND: BT, formerly British Telecom, is the leading supplier of local, national and international phone and data services in the United Kingdom. A publicly owned company with full-year sales of $29.39 billion, it has established a powerful family of alliances and joint ventures in every country in the world.

PLATFORMS: Digital Alpha Server running on UNIX with an Oracle7.1 database.

PROBLEM SOLVED: BT sales and marketing needed to target customers for its Business Highway direct mail campaign which launched in September 1998. Our assignment was to develop profiles of customers with the greatest propensity to purchase and the highest likely comparative value. These customer profiles were based on response and transactional data for the mailing, plus existing customer, billing, product and historical data. They would be used in future Business Highway promotions to improve response rate and increase revenues. We used Clementine 5.0's exploratory data and characteristics analysis capabilities ­ particularly decision tree analysis ­ to examine individual data attributes, such as geographic location and number of phone lines. With Clementine, we were able to evaluate relevant attributes so quickly that we developed satisfactory initial selection target criteria even before completing the actual models. Current results show that our models have improved the response rate to the latest Business Highway campaign by 50 percent.

PRODUCT FUNCTIONALITY: Developing accurate customer profiles requires a sophisticated data mining product such as Clementine which can sift through data from a variety of sources and discover patterns and trends. Clementine's interactive stream approach to data mining enabled us to analyze and visualize data and then apply several different algorithms to build a model. We will continue to use Clementine for numerous profiling projects that will assist sales and marketing. Because Clementine is easy to customize, we also plan to add source code with specific functionality developed by our own data mining experts.

STRENGTHS: Clementine's greatest strength is its ease of use for exploratory analysis and visualization of data. It enables you to discover quickly whether a line of investigation will be worthwhile, thus reducing the cost of failure. In only a few days, you can construct a series of explanatory models based on different algorithms and determine their predictive ability.

WEAKNESSES: The main weakness of Clementine is its simulated Windows front end which makes the delivery of graphics to the screen slow and cumbersome. As a result, Clementine can be used adequately on the machine on which it is installed but is difficult to use remotely over phone lines.

SELECTION CRITERIA: We selected Clementine as our data mining software of choice because it provides a rich set of analytics, including neural networks, graphs, rule induction models, clustering methods and linear regression models, at a highly reasonable cost. Although we evaluated another product with many modeling features, its yearly maintenance cost was higher than the purchase price of Clementine. We believe our cost savings achieved in the Business Highway campaign alone will pay for Clementine quite easily.

DELIVERABLES: Clementine's main output is insights about data and visual representations of those insights through graphs and models. The results of our Clementine analyses were customer profiles based on models. Specifically, to our customers in sales and marketing, we delivered lists of clients who should be targeted by the Business Highway campaign and charts showing why these customers would prove to be profitable.

VENDOR SUPPORT: Our dealings with SPSS support were limited because we encountered few problems with the product, before or after implementation. On the few occasions we called support, our calls were returned by knowledgeable people who quickly resolved our problem. However, the response was sometimes a little slow ­ i.e., calls were returned the next day, not the same day.

DOCUMENTATION: We found the documentation to be quite satisfactory, although rarely required, because Clementine's interface intuitively guides you through a logical progression in analysis. We consulted the documentation half-a-dozen times, but we could have run the product even without it.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access