Microsoft Focuses on New Customer Groups
About four years ago, Microsoft's midmarket customer development team was charged with deploying direct marketing programs to reach its customers, companies ranging from 50 to 500 employees.
The team enlisted Loyalty Builders, a Portsmouth, NH, relationship marketing company, along with another internal Microsoft team and advertising agencies to develop campaign material.
To reach these customers, Microsoft first collects transaction data from past purchases and interactions, then uses loyalty analysis with Loyalty Builders to predict buying behavior, such as which customers would most likely make repeat purchases. It then develops and deploys appropriate marketing campaigns, such as direct mail, e-mail or telesales.
For loyalty analysis, Loyalty Builders uses a Finite State Machine approach, which is "a mathematical term for building a specific model for every single customer," said Mark Klein, CEO/manager, Loyalty Builders. "With this approach, you can include a wide variety of input variables. You are not just limited to transaction data."
Klein said this analysis is more effective than traditional regression or RFM techniques. "Why is the fact that someone purchased recently the best predictor of loyalty?" Klein asked.
Using this technique, Microsoft learned in one example that the best customers to turn to for revenue growth were a group Loyalty Builders labeled "bad, but good lately." They were essentially newer Microsoft customers in lower loyalty groups but with increasing buying rates. This group had the greatest potential for revenue growth for Microsoft.
Microsoft still focuses on its best customers, "but we don't have to invest as much money in reaching that group," said Jennifer Dorsey, customer loyalty and development manager, Microsoft. "Just targeting the high-ranking customers leaves a lot of potential revenue on the table."
In general, Dorsey said, these types of modeled lists "outperform our non-modeled lists. We get a response rate that is four times higher [than non-modeled lists]."
In addition, she said the technique increases the number of qualified opportunities, and the cost-per-response decreases.
Klein said that Microsoft is eager to experiment and test programs with various customer groups, "as opposed to always looking just at the very top customers."