Baylor Makes the Grade With Recruitment Analysis
"We take 15 to 20 different variables we might use for prediction purposes, such as the prospective students' test scores, whether their parents are alumni or not, their extracurricular activities, their distance from Baylor, whether they have ever attended an event at Baylor and whether they contacted us or we contacted them, and run them through Enterprise Miner," said Dr. Tom Bohannon, assistant vice president and director of information management and testing services at Baylor. "Then ... we come out with about seven predictive variables that we use to identify those students with high scores."
Baylor, Waco, TX, has more than 13,000 students and is the oldest college in Texas. It has contact with 80,000 prospective students yearly. Names come from inquiries by prospective students via telephone calls, e-mail and mail, and from prospecting mailing lists. Names also come from other sources, such as when high school students request that their Scholastic Achievement Test scores be sent to Baylor.
Baylor has used Enterprise Miner from SAS Institute Inc., Cary, NC, for three years to evaluate prospective students. The software offers decision trees, clustering, neural networks, data-mining regression and other algorithms in a single package. The university then develops a prediction model based on information about students "from which we get a probability score that shows the probability that they will enroll at Baylor," Bohannon said. "If a student receives a score of 0.9, it is basically saying that the probability is 90 percent that the student will enroll in Baylor."
The university uses the score to decide what type of mailing piece a student receives.
"If a student received a high score, they would receive a very expensive marketing brochure about Baylor," Bohannon said. "Based on budget constraints, however, we would probably not send that type of marketing brochure to those students who scored very low because it is not very likely that the student is really very interested in Baylor."
For students with lower scores, Baylor "sends them a postcard identifying the materials that we sent out to the students who achieved a high score and asks them if they would like a copy of it." Then, Bohannon said, "if they send the card back in, that goes into the model, and that is obviously going to raise the productivity score the next time we run the model."
Enterprise Miner helps Baylor meet its enrollment goals, Bohannon said, though the school fell short of its goal this fall.
"I think there were a number of reasons for this, such as the economy and 9/11, but we are hoping we will meet our goals next year," he said.
Previously, the university has met its enrollment goals with the program, though Bohannon would not give specifics about recruitment efforts.
The university also plans to use data mining and data analytics for fundraising in the future, and it introduced a master's program for CRM at its Hankamer School of Business in fall 2001 "in response to a growing need in the business world for people who can effectively manage customer relations," Bohannon said.
The program CRM specialization consists of five courses plus internships. The courses include data warehousing and data mining, introduction to CRM, CRM strategy and implementation, and CRM analysis. Internally, the program works closely with the business school's new Center for Data Mining and Data Warehousing, The program also uses relationships with organizations such as SAS, American Airlines, PwC, Oracle, Teradata and Pizza Hut. The program had 12 students last year and eight this year.