Last year, San Francisco-based StubHub Inc. began a major overhaul of its BI systems, ditching Microsoft Access and Excel in favor of enterprise-class best-of-breed tools, according to Rob Singer, the company's director of business intelligence. StubHub provides a secondary marketplace for tickets to sporting, concert, theater and other live entertainment events -- listing everything from baseball tickets to nightclub shows. StubHub doesn't actually sell tickets, it facilitates transactions between buyers and sellers via its Web site, making money from a percentage of each sale. StubHub calls it a "fan to fan marketplace," Singer said.
"We're not a retail model -- we're really an information company. Our business is about data quality and data management," Singer said. "As we grow and more tickets are getting posted on the site, the data we have about our business is growing at roughly the same rate."
StubHub needed to be savvier with its information assets and wanted to increase its sales and marketing through multiple media channels, Singer said. The company realized its data was a potential goldmine of insight about buyers, sellers and ticket market trends. Getting to that insight was nearly impossible, however, because of the database size limitations and rudimentary features of the Microsoft tools. For a company regularly tripling in size, with more than $200 million in revenue last year, Access simply wouldn't scale. Neither would the BI team, Singer pointed out.
"From an information consumption standpoint, we needed to go from a request model to a self-service model so our analysts could focus on the big fish, not just report creation," Singer said.
The BI team began evaluating systems, selecting Oracle for its data warehouse; Orem, Utah-based Omniture Inc. for Web analytics; and, Cary, N.C.-based SAS Institute for "deep analytics," Singer said. The BI tool decision came down to Cognos Inc., based in Ottawa, versus BusinessObjects SA, with dual headquarters in San Jose, Calif. and Paris. The team set up both tools in a test environment, using samples of StubHub's own data. Internal business users evaluated reporting features and interfaces, while the technical team focused on integration. StubHub chose BusinessObjects and cited its technical approach as a good fit for its needs, Singer said. Choosing the tool wasn't the hard part.
The team is following a "concurrent phase model" to roll out the system, Singer said. The new data warehouse went live in December, and as components stabilize, the team layers on new features. Enterprise reporting was rolled out in January, Web analytics went online in February, and there's a roadmap for the coming year. The team plans to deploy dashboards this year, and in the longer term, a near-time environment and more customer-facing BI. Integrating the technology isn't the hardest part, either, Singer said.
"The biggest challenge is in changing mind-set, when people have been working for years with Excel pivot tables," Singer said. "People get very used to looking at data from a certain perspective."
However, new BI users are seeing the light, he said. The product development team scrutinized data about frequent ticket sellers and optimized processes for these profitable customers. The company has gained a better understanding of its multiple marketing channels and how Web, telesales, radio and other advertising methods work together. Marketing has new customer insight from Web-based analytics and can better target e-mail campaigns. User requests for new reports from the BI team have slowed dramatically because users can now create their own reports, using a drag-and-drop, self-service reporting interface.
Managing user expectations is critical in a long-term BI project, Singer advised. Set project timelines up front and then double them, he said, to ensure that the team has enough leeway to implement the system the right way the first time. Prior to this project, StubHub users hadn't been used to the implementation process of an enterprise-class system that can take a year or more to fully roll out, Singer said. The team had to constantly "sell" the concept and long-term benefits.
Most important, set aside time and resources for user training and education, Singer added. It takes time for an organization to understand the value of enterprise level BI. Don't expect users to be excited about getting rid of Excel or other familiar legacy tools, he cautioned.
"People are used to seeing things a certain way. They have that inherent expectation of replication. You want to debunk that theory as quickly as possible. There's really no value in reproducing what was already there," Singer concluded, "The key is to focus on why [BI] is the right move."