From the November 2010 issue of Treasury & Risk magazine

Bronze AHA Winner in Retirement & Benefits


The enterprise risk management group at Paychex, a provider of payroll and human resources services, used its abilities in modeling and predictive analytics to come up with metrics that identify the companies already using its payroll services that are most likely to purchase retirement plan services within the next 12 months.

The risk management department's modeling group looked at 400 different pieces of data the company has about each of its customers and employed regression analysis to select the 17 metrics most predictive of whether a company would be in the market for 401(k) record-keeping.

The 17 variables that it identified include the length of time a company has been a Paychex customer, the percentage of employees using direct deposit and the frequency with which the company deposits federal tax payments. "It's the combination of all these different characteristics and their weightings put together that raises that customer up to ready to buy our product," says Frank Fiorille, director of enterprise risk management at Paychex.

It took four months to put together the model, which is called Predictive Algorithm Targeting Retirement Investment Clients Knowledge, or PATRICK. The model has proved successful in boosting retirement services sales by identifying the best customers to target. While the ratio between telemarketing calls and appointments scheduled had been 26 to 1, the ratio for calls to customers identified by PATRICK is 18 to 1. And while Paychex typically closes a retirement services sale at 11.7% of scheduled meetings, that rate improves to 14% for customers identified by PATRICK.

Fiorille notes that risk management teams traditionally focus on what bad things could happen to the company. "What we did, and this is where risk management is going right now," he says, "we moved outside of that traditional domain to focus on the upside of risk."

Comments

Advertisement. Closing in 15 seconds.