Mercer Consumer - Best Use of Technology in Sales
Company: Mercer Consumer, Des Moines, IA
Company Description: Mercer is a global consulting leader in talent, health, retirements and investments. Mercer helps clients around the world advance the health, wealth and performance of their most vital asset—their people. Mercer's 20,500+ employees are based in more than 40 countries. Mercer is a wholly subsidiary of Marsh & McLennan Companies, a global team of professional services companies.
Nomination Category: Sales Awards Achievement Categories
Nomination Sub Category: Best Use of Technology in Sales
Nomination Title: Short-term Recovery Predictive Model
Tell the story about how your organization used technology to improve sales operations, drive sales growth, and/or improve customer satisfaction since the beginning of July 2016 (up to 650 words). Focus on specific accomplishments, and relate these accomplishments to past performance or industry norms. Be sure to mention obstacles overcome, innovations or discoveries made, and outcomes:
In marketing, the essence behind building predictive response models is determining whom to solicit with your materials with a high degree of confidence. Accomplishing this maximizes your rate of return and minimizes your production and postal expenses. Developing an effective model gives the marketer a huge advantage over the competition. But if it were easy, everyone could do it.
Mercer Consumer’s Analytics team developed such a model for one client, and the result promises to extend the benefits to the entire organization. Here’s what the model helped achieve in its first use, a mailing for a supplemental medical insurance product:
Increased the paid response rate by 44%
Decreased the number of pieces mailed and cost per premium ratio by 32%
One particular association client was chosen as the target audience for the model’s development and testing. This client was selected because of its unusual attributes. The association is considered part of our military groups and is generally included on mailings our other military associations receive. However, this group is made up of civilians who have different insurance needs and traits than former and retired military men and women. Our goal was to understand how to differentiate those members from the rest of our military groups.
To craft the new model, the Mercer Analytics team applied a keen eye to the tools at hand: the association solicitation, membership, and response data available in our proprietary Mercer Data Warehouse, augmented with purchased geographic, demographic and behavioral fields from Merkle, one of the nation’s premier performance marketing agencies. The team then data mined and analyzed this complex data and built the final predictive response model using SAS®, a software suite for advanced analytics.
First, the team analyzed response rates from various sub-groups based on attributes such as marital status and children at home — among the hundreds of pieces of data available. But raw data is like recipe ingredients: You have to know how to combine them, and in what quantities. You also have to weigh the data points for importance through a technique known as logistic regression while ensuring they are not significantly correlated with one another, so you can accurately ascertain which factors matter most. Finally, you have to be as certain as can be that the factors accurately represent the population the model is trying to predict. The analysts strove for at least 95% confidence.
After sifting through the myriad records with all that in mind, the team scored each of the 100,000-plus client members individually with the final predictive response model. The top 10 percent were the most likely to respond, and the bottom 10 percent the least.