There\u2019s no shortage of technology at work when you call a customer service line, including customer relationship management systems and voice-recognition programs. From the second you dial that toll-free number until the moment the agent answers, a slew of tools is whirring in the background to make the interaction as helpful as possible.\n \n But anyone who\u2019s endured a less-than-productive 1-800 experience knows these tools don\u2019t always get you to the right person\u2014let alone the best person\u2014right away.\n \n \u201cIt\u2019s a thousand-mile journey between when a customer picks up the phone and when they land on a customer service representative\u2019s desktop,\u201d says Cameron Hurst, a 15-year veteran of call-center-technology management. \u201cThe problem is, we\u2019ve overlooked the last 500 feet.\u201d\n \n Assurant Solutions, where Hurst is vice president of targeted solutions, provides specialized insurance products from its base in Atlanta, and it has developed a better way to match customers and call center agents using analytics. Most call centers are heavy users of rules-based analytics, such as systems that route calls based on \u201cif-then\u201d formulas. (If your account balance is greater than $10,000, then you are sent to a premier agent. If the call is made between 8 a.m. and 5 p.m., you\u2019re routed to Boise instead of Bangalore.)\n \n This wasn\u2019t enough for Assurant Solutions, which sells and supports debt-cancellation products, credit card insurance, and other financial-protection products through Fortune 50 financial institutions. The company has a high acquisition rate\u2014most of its products come with a trial period\u2014and a high customer churn rate. Seven years ago, it retained just 16 percent of its customers.\n \n So it turned to model-based analytics. The company created a proprietary system that analyzed the attributes of all calls over six months to create models of the successful ones. It then examined a host of customer variables\u2014such as age, account balance, credit-to-balance ratio, and persistence in calling\u2014to create models of different customer \u201cclusters.\u201d \u201cThe models are reduced to algorithms [which become] Java code, and that code powers a matching engine,\u201d Hurst explains. \u201cIt scans thousands of agents and finds that one optimal match. It\u2019s kind of like eHarmony\u2014there\u2019s someone for everyone.\u201d Unlike most matching systems, Assurant\u2019s looks at agents who are already on a call\u2014not just those who are available\u2014and determines if they will be wrapping up in time to deal with their next customer soul mate.\n \n Since implementing the system, Assurant\u2019s customer-retention rate has jumped 190 percent to around 47 percent. Most customer-service-related technologies promise a 10 percent to 25 percent improvement in customer retention, says Hurst, who was formerly a CIO at a large community bank.\n \n The system and its processes\u2014now patented\u2014showed enough promise that IBM partnered with Assurant to create a tool called the Real-Time Analytics Matching Platform (RAMP) which it began selling earlier this year. Assurant signed a revenue-sharing agreement with Big Blue for future RAMP sales and became the system\u2019s first buyer.\n \n Hurst expects to see a lot of interest, particularly among financial services companies desperate to \u201csqueeze the most juice out of the lemon\u201d with their customers. \u201cThey\u2019re deep into analytics,\u201d says Hurst. \u201cBut this is the one thing they\u2019re missing that\u2019s very, very important.\u201d\n Stephanie Overby is a freelance writer based in Massachusetts.