Increasing customer attrition jolted management at Assurant to investigate its contact center operations. Although the contact center appeared to be finely tuned, after deeply analyzing customer data, the specialty insurance company discovered missed opportunities to establish more personalized relationships between its agents and customers.
"We were well managed, well run, and had a seasoned staff, but we were still struggling with turnover issues," says Cameron Hurst, vice president of targeted solutions at Assurant. "What the [statisticians] were able to discover was that there are attributes that [would] enable us to connect a customer with a specific CSR when he called in."
In response to the findings, Assurant developed a home-grown agent call-matching technology called Real-Time Analytics Matching Platform (RAMP). (IBM acquired RAMP last year.) Assurant customized the solution to analyze a combination of key attributes like customers' customers' persistency, the CSRs' tenure and ability to save a customer, and the financial metrics around the customers, such as their future economic value and the potential revenue that they could generate. The technology then matches CSRs with customers based on these algorithms.
Hurst reports that this practice elevated the customer save rate from 15.5 to 23.5 percent in the first year after deploying the solution. The save fee rate (the CSRs' ability to save customers' eligible fees) increased from 16.5 percent to 36 percent in the first year, and the 2010 save fee rate was 47 percent. "On our best days we see a 57 percent save rate," Hurst says. What's a save fee rate?
In addition, agent attrition dropped 25 percent that first year, which Hurst attributes to placing CSRs in interactions that make them successful and ultimately earn them more money. "Happier interactions result in happier CSRs," he says.
Assurant is proof that assigning calls to agents based on a combination of specific customer characteristics and agent performance is effective in boosting sales and retaining customers and employees. While still in its infancy, the ability to match a customer with the agent who has the best chance of building a rapport with him has numerous benefits.
In fact, according to Andrew Kokes, Sitel's vice president of global product management, such in-depth agent-customer matching leads to improved sales, customer satisfaction, and shorter handle times. As a result, he sees growing interest among companies wanting to use what he calls intelligent pairing. "I see it as an emerging trend," he says. "It will take a little while, but there are some early adoptersbut I do see it becoming a much more popular approach."
While its benefits are easily understood, finding the optimal match between caller and agent is an in-depth and evolving process based on both customer and agent characteristics that can range from gender and age to education and location.
Kokes cautions that matching technology is not something a company can drop in and turn on. "There's a significant amount of customization. To get the agent information and to get the call outcome data, you need to pull from [various] systems and configure with those," he says.
Karen Hardy, senior product marketing manager at Alcatel-Lucent, recommends developing a roll-out plan as the first step in creating a multivariable agent-customer matching program. "If you try to shift everything to a granular level overnight, it will probably create more complexity," she says.
According to Hardy, some companies that are finding success in their contact center matching strategies give online psychographic tests to agents, record interactions with customers, and combine that information with psychographic data to better understand and predict optimal matches. "To be able to refine and mange the [customer-agent] profile has become key," she says. "We see this model transforming the way in which businesses operate, using it to ensure that they [have] the right leaders in place, managing the right staff, and keeping the staff engaged at all times."
Toby Cook, customer analytics practice leader at IBM's global business services division, recommends analyzing historical customer data, as well as the performance of the CSR, when matching agents and customers. Psychographic data can also be helpful in this process, but is not always necessary. "We're finding that it's not as important to the match as some other things," Cook says. "If you're calling to cancel debt protection services for a credit card, [for example], then psychographic data can be helpful."
Finding the right match at UPMC
The University of Pittsburgh Medical Center (UPMC) relies on geographic data to match its members to their personal concierges, a program that UPMC offers to its 78,000 senior members on Medicaid. When members calls UPMC's contact center, the Genesys system it uses automatically matches them with their personal concierge.
Anne Palmerine, director of member and enrollment services at UPMC, says that the member and concierge are matched on geographic variables. When the member calls in, he provides his member ID. The agents are grouped in pods of five, so if a particular agent is unavailable, then the call is automatically rerouted to another agent in that group who may be familiar with the member's issue. "The idea is that we make an effort to get them connected to their personal concierge," Palmerine says.
Additionally, UPMC ensures that the concierges are continuing the personal relationships when they conduct outbound calls to their assigned members for such customer-focused initiatives as calling them on their birthdays, alerting members of changes in premiums, educating them on benefits, helping to schedule appointments, and even taking preventative care measures such as letting them know whether they're eligible for a glaucoma screening.
Palmerine reports positive results since deploying the agent-matching strategy. The organization is in the midst of open enrollment and it expects to see 15 to 20 percent growth over last year, and customer satisfaction already has increased from 89 percent to 91.8 percent.
Next, Palmerine says that UPMC plans to match its chat agents with members. She says, "We continue to refine and ask, 'What else can we do to make sure that we are establishing a one-to-one relationship?'"