In less than one year, this fabled wagering powerhouse developed advanced CRM and predictive analytics for deeper customer insightIn less than one year, this fabled wagering powerhouse developed advanced CRM and predictive analytics for deeper customer insight

Field Report: Churchill Downs, Louisville, Ky

"This is a triple crown race," says Atique Shah, vice president of CRM and technology solutions at Churchill Downs. Shah is the jockey, so to speak, of a significant implementation of predictive analytics and advanced CRM. The setting for such cutting-edge customer intelligence is fairly surprising; Churchill Downs is most famous for the Kentucky Derby, since 1875 the great annual festival of thoroughbred horse racing held in Louisville, Ky.

It would be hard to imagine anything more tradition-bound than the Kentucky Derby. However, the company has expanded aggressively over the past 10 years to include several other tracks, a casino and a network of off-track betting, simulcast racing operations. Churchill Downs claims a market share of 20 percent of all thoroughbred wagering in the United States.

Shah arrived in Louisville in December 2003 after an award-winning stint at GSI Commerce, where he led a multichannel CRM initiative that included database marketing and data mining. His mission was to employ CRM analytics to extend the influence of Churchill Downs' core values of strong customer service and develop one-to-one relationships that would be integral to its brand across all channels and properties.

One of his first moves was to bring in E.piphany's CRM suite to create a software platform from which he could upgrade the company's marketing effectiveness. Then, to build customer intelligence, Shah turned to the data universe in front of him: more than 100 different data sources, all of which were critical to reaching the lofty goal of a 360-degree view of the customer. Sources included relational databases, Microsoft Access and Excel, flat files and more. Shah's team consolidated data into a primary operational data store and an enterprise data warehouse running on Oracle 9i.

Dismissing prognostications from a prominent systems integrator that it would take three to five years, Shah hired a crack team of 18 CRM experts and worked with Ascential, E.piphany and SPSS services to finish development in November 2004. "We loaded seven data sources into our CRM system initially, choosing the most valuable from transaction, lifestyle and demographic perspectives."

Continuously adding data sources would prove important because Shah's real target was a platform for predictive analytics — "the foundation of everything we believe in." Employing SPSS's Clementine 9.0 data mining workbench, Churchill Downs built models that can, with increasing accuracy, predict the future potential of a customer based on one day's worth of transactions. Then came the confrontation with tradition. Churchill Downs has a longstanding loyalty rewards program backed by a database of some 200 million transactions by hundreds of thousands of customers. The company had long ago decided on four customer tiers, but Shah felt it was time for the data to tell how many tiers, or segments, really made sense. "Without imposing anything on the data, Clementine said there were nine segments."

Rolling out the system to the businesspeople, Shah's team had to explain why nine segments based on multiple attributes should be "the foundation for every program we put in place — all the direct mail, call center, ticket sales and everything." There was resistance, but compromise was reached. "Now, we think in the segment, but we act in the tier. We don't go to the customer and say, 'Hey, you're a Smarty Steve'-one of our top customer segments. We still use the tiers."

Smarty Steves win 80 percent of the time, Shah says. "In horse racing, data is God. The more information you have about the jockey, horse, track conditions and so on, the better your chances of winning. And I want customers to win: In horse racing, you're not betting against the house; you're betting against everybody else betting in that race. I'd rather they win because then they come back and play more."

Predictive analytics have had a pervasive effect beyond strictly CRM. "Because of our data knowledge, we're implementing a brand-new ticketing system, Web site and content management system," says Shah. "Without casting out tradition, we're doing things that we've never done before."