Afni deploys predictive analytics to drive million-dollar financial benefits

Using a smarter approach to debt recovery to identify the best payers and focus collection efforts

Published on 17-Nov-2013

"Without a doubt, the predictive analytics capabilities provided by IBM SPSS Modeler are a key differentiator for our business." - Barry Gamage, Director of Advanced Analytics, Afni


Professional Services

Deployment country:
United States

SPSS Modeler, BA - Business Analytics, BA - Risk Analytics, BA - Predictive Analytics


Headquartered in Bloomington, IL, Afni employs approximately 4,000 people. With expertise in a wide range of industries including communications, insurance and healthcare, the company delivers a comprehensive set of offerings that includes acquisition and enrollment; care and loyalty; up-sell and cross-sell; and receivables and subrogation services.

Business need:
Afni wanted to offer more profitable debt-recovery services to its clients by accurately and cost-efficiently predicting which debtors were most likely to pay, and prioritizing collection efforts accordingly.

The company deployed IBM® SPSS® Modeler and IBM SPSS Modeler Server software in-house – replacing third-party analytics services and enabling the company to develop more advanced predictive models.

Helps boost collections success rates by up to 15 percent. Delivers immediate six-figure financial benefits to Afni’s clients.

Enables greater investment in analytics to further improve success rates.

Case Study

How can companies tasked with recovering consumer debt for clients make best use of resources? The key is to focus collection strategies on the people who are most likely to pay – but how can you tell the good debtors from the bad ones?

This was the challenge facing Afni, a leading provider of contact center outsourcing for enterprises in the United States and the Philippines. Working together with IBM, the company implemented IBM SPSS Modeler – using predictive analytics to deliver up to 94 percent reductions in manual processing, six-figure financial benefits for its receivables business, and an estimated potential return of several million dollars across the entire company.

Corporate background
Headquartered in Bloomington, IL, Afni employs approximately 4,000 people. With expertise in a wide range of industries including communications, insurance and healthcare, the company delivers a comprehensive set of offerings that includes acquisition and enrollment; care and loyalty; up-sell and cross-sell; and receivables and subrogation services.

Detect good payers to boost profitability
To enable maximum cost-efficiency in its receivables and subrogation business, Afni wanted to focus its collection efforts on the debtors who have the highest probability of repaying.

As Barry Gamage, Director of Advanced Analytics, explains, “Afni provides late-stage, post charge-off debt collection services for many clients. Only a small proportion of the debtors in these portfolios repay their creditors, and each time we accept a package of receivables from a client, we take on a degree of business risk.

“To maximize returns for our clients and minimize our risk, we use predictive analytics to estimate each debtor’s individual circumstances and give them a score based on their probability of repaying.”

Afni knew that building smarter analytics models would help improve the accuracy of its predictions – enabling it to spend more on high-probability payers, and gain a greater return for each dollar spent on telephone calls, letters and “skip-tracing” (the process of locating a debtor’s current whereabouts if their address details are missing or inaccurate).

“In the past, we used a third-party provider for predictive analytics, which was expensive,” says Gamage. “We calculated that making changes to our models and running multiple scorecards per day would result in prohibitive costs if we continued with this approach. We were ready to develop an in-house predictive analytics capability, and we wanted to do it quickly.”

Choosing a smarter approach to predictive analytics
After evaluating a number of analytics solutions from different vendors based on criteria including ease of use, functionality and total cost of ownership, Afni chose IBM SPSS Modeler.

“Of all the solutions we considered, we felt that IBM SPSS Modeler was the best fit for our business requirements,” says Gamage. “We sought a short implementation timeline, and were confident the IBM solution offered the user-friendly, cost-efficient package of advanced analytics functionalities we needed to achieve our business goals.”

He adds: “Once we had built the business case for an in-house predictive analytics solution, there were no difficulties securing senior buy-in; our executives immediately understood the need for investment in this area.”

With the full backing of its executive team, Afni deployed IBM SPSS Modeler and IBM SPSS Modeler Server software.

“Implementing IBM SPSS Modeler was a straightforward process, and the solution went live on time and within budget,” says Gamage. “Once the installation was complete, we participated in some live online training sessions with IBM. These sessions were extremely useful, and helped us to get up to speed with the functionalities of IBM SPSS Modeler rapidly.”

Delivering game-changing business insights
Today, Afni’s IBM SPSS Modeler solution forms the heart of its advanced analytics department. By augmenting its predictive analytics capabilities, the company has achieved its goal of boosting profitability for its clients and itself, and minimizing risk in its receivables business.

“Before the IBM solution, we had to make do with predictions based on a limited set of data,” says Gamage. “Because our in-house analytics program is more cost-effective, we can now enrich our data with credit reports, ZIP+4 information from the census bureau, average incomes, house values, and levels of education in debtors’ neighborhoods.”

Using automatic segmentation tools in IBM SPSS Modeler, Afni can now identify the strongest predictors of a good or bad payer – avoiding time consuming trial-and-error processes. Based on these indicators, the solution automatically scores each debtor with ultra-low, low, medium, high or ultra-high probability of repayment. These classifications determine the threshold for the maximum number of telephone calls, letters and skip-traces that Afni will invest in for each debtor’s individual treatment plan. Depending on each individual score, the solution automatically initiates automated mailing or dialing campaigns – streamlining the contact process.

Automatically identifying the best payers
“We had not anticipated just how accurate ZIP code predictors could be,” says Gamage. “The SPSS Modeler solution has revealed that neighborhoods with higher than average incomes and house values are strong indicators of good payers – enabling us to focus our efforts on cases with the highest probability of returns. In fact, we estimate that the solution will deliver six-figure savings for our receivables business and a return of up to 15 percent for our clients.”

By bringing predictive analytics in-house, Afni can score debt files multiple times over the course of their treatment plans – which would have been prohibitively expensive before. If a debtor’s score changes over time, Afni refines its contact strategies based on up-to-date information.

Increasing success rates for subrogation
The predictive analytics capability is also enabling the company to improve debt-recovery outcomes in its subrogation business – which acts on behalf of insurers to recover the cost of claims from the party who is at fault or their insurance company.

“Subrogation is an important part of our service portfolio,” says Gamage. “In cases of car accidents, we are often contracted to subrogate an insurance claim from either the at-fault party or the at-fault party’s insurer.

“In the past, pursuing a subrogation claim against an uninsured driver had a low probability of success. Since we have no legal right to see credit data in such circumstances, it is difficult to identify the drivers who are most and least likely to pay.

“With IBM SPSS Modeler, we can now use ZIP+4 data to enhance our insight into uninsured drivers, which helps predict good payers more accurately. As a result, we estimate that we can increase successful subrogation collections by as much as 15 percent – helping us recover more dollars for our clients.”

Cutting manual processing by 94 percent
Using IBM SPSS Modeler also enables the company to accelerate the process of identifying opportunities to pursue subrogation claims that might previously have been overlooked.

Gamage comments: “Clients sometimes ask us to perform closed-file reviews, during which we go through tens of thousands of car-accident records to catch potential subrogation opportunities that may previously have slipped through the cracks. Previously, determining which driver was at fault meant reading through each record individually – generating hours upon hours of manual work.

“Today, we have developed a solution called ACER – Automated Closed-file Evaluation and Review. It uses text analytics in SPSS Modeler to mine closed files for key indicators of at-fault drivers, and flag only the most likely candidates for manual review by our subrogation team.

“In one recent project, we used ACER to cut our manual closed-file review workload down from 100,000 to just 6,000 records – a 94 percent reduction in manual effort. We found the same number of subrogation claims to pursue as a full manual review would have yielded – but at a much lower cost to our client.”

Potential for multi-million dollar savings
The successful predictive analytics projects described in this case study are just the first of many that Afni hopes to move into production. The advanced analytics team currently has a list of 108 projects on its agenda, which are estimated to deliver total cost benefits of several million dollars if they all reach implementation. These will improve results for clients and give Afni more ability to invest in research and development of its service set.

Gamage concludes: “Without a doubt, the predictive analytics capabilities provided by IBM SPSS Modeler are a key differentiator for our business. Thanks to the IBM solution, we can focus our investment and resources on people who are most likely to repay their debts – helping us to deliver even greater returns for our clients.”

About IBM Business Analytics
IBM Business Analytics software delivers data-driven insights that help organizations work smarter and outperform their peers. This comprehensive portfolio includes solutions for business intelligence, predictive analytics and decision management, performance management, and risk management.

Business Analytics solutions enable companies to identify and visualize trends and patterns in areas, such as customer analytics, that can have a profound effect on business performance. They can compare scenarios, anticipate potential threats and opportunities, better plan, budget and forecast resources, balance risks against expected returns and work to meet regulatory requirements. By making analytics widely available, organizations can align tactical and strategic decision-making to achieve business goals.

For more information
For further information please visit

Products and services used

IBM products and services that were used in this case study.

SPSS Modeler Server, SPSS Modeler

Legal Information

© Copyright IBM Corporation 2013. IBM Corporation, Software Group, Route 100, Somers, NY 10589. Produced in the United States. October 2013. IBM, the IBM logo,, and SPSS are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the web at: This document is current as of the initial date of publication and may be changed by IBM at any time. Not all offerings are available in every country in which IBM operates. The client examples cited are presented for illustrative purposes only. Actual performance results may vary depending on specific configurations and operating conditions. THE INFORMATION IN THIS DOCUMENT IS PROVIDED “AS IS” WITHOUT ANY WARRANTY, EXPRESS OR IMPLIED, INCLUDING WITHOUT ANY WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND ANY WARRANTY OR CONDITION OF NON-INFRINGEMENT. IBM products are warranted according to the terms and conditions of the agreements under which they are provided. The client is responsible for ensuring compliance with laws and regulations applicable to it. IBM does not provide legal advice or represent or warrant that its services or products will ensure that the client is in compliance with any law or regulation.