Doggedly pursuing a business answer

LONDON, England (CNN) -- It's a common dilemma for a start-up business -- to expand, you need complex marketing data analysis, but this is an expensive process.
One company's solution? Go back to school.
Petplan USA, which provides insurance for pet owners to cover veterinary bills and the like, was itself the product of a business competition at the University of Pennsylvania's Wharton School in 2003.
Founders and Wharton MBAs, Chris and Natasha Ashton, thus had the perfect place to go for advice when they realized their fledgling firm could not afford to pay for a mainstream market research group to help devise a possible nationwide marketing strategy.
They were put in touch with Alan Abrahams, a visiting Wharton professor who teaches the complex and expertise-heavy subject of data mining, in which large amounts of information are surveyed to glean the required information.
After a test run in the summer of 2006, in which a student dug out some sample data about the location of veterinary surgeons' offices nationwide, the co-operation went further.
Abrahams used the Astons' dilemma as a "live case" for his data-mining class, instructing students to come up with the best possible algorithms for analyzing the reams of data relevant to Petplan's marketing campaign.
Surprise findings
First, the Ashtons visited Abrahams' class, presented their business plan and took questions. "The students pushed back on our ideas and really tested our assumptions," Chris Ashton said.
The three best plans were then used to help Petplan decide where to focus its marketing efforts.
"The project would help us to identify demographically favorable parts of the country," Chris Ashton explained. "The students were trying to identify indicators that would make households more likely to buy pet insurance."
The analyses came up with some interesting findings -- not only did the students confirm that average household income and average age played a big role in deciding who was most inclined to buy pet insurance, they also discovered that proximity to pet superstores and even race played an apparent role.
As well as providing very real benefits to a business, the exercise helped get the students out of the theoretical forum of the classroom and textbook and into the real world.
"In most, if not all, of my classes, you go to the lectures and do readings," said Daniel Sabido, who came up with one of the three best solutions.
"Then it's a lot of applying what you learned in a test. This class was a lot more practical. This sort of setup should be done again. It was very useful for us as students."
The exercise was also useful in that the data mining class includes both business students and those from the university's engineering department, among them graduate computing student Rosskyn Dsouza, who gained some new commercial insights.
"I learned how to calculate the cost-to-market of Petplan's policies and how to calculate the profits that Petplan or any insurance company could make by selling their product," he said.