IVR (Interactive Voice Response) is perhaps the most widely used call center technology worldwide nowadays. It is basically a communications system that accepts a combination of voice and telephone touch-tone keypad input and provides appropriate responses in the form of voice or other media.

IVRs were first launched almost a decade ago to enable customers to "self-service" a large proportion of their requirements through an automated system. The key reason for introducing this technology was to automate selected tasks that customers needed to do over the phone, freeing up call center agents for handling more complex and convoluted tasks. Some research indicated that every second saved on the average length of a call to an agent meant a saving of $120,000 a year. Where call center agents numbered in the thousands, savings per second was assessed to have been at many millions of dollars per year! No small amount!! The math led every one to believe this was the future of call centers.

In addition, it would also enable businesses to provide round-the-clock services to their customers as well as manage peak call volumes without necessarily having a 24x7 shift for its call center, thereby leading to cost effective management of calls.

The buzz around this technology at that time was so high that many predicted that the end of call centers was near. In the frenzy of automation, this technology was adopted widely across industries, geographies and businesses.

However, over time, IVR earned a bad reputation as it resulted in poor quality of interactions with customers and lower resolution of calls, resulting in customer abandonment of IVR calls. At the first instance an opportunity was offered to them to talk to a live agent. This resulted in neutralizing all IVR investments and increasing call center agent costs. The businesses realized the need to enhance usability of IVR through customer friendly designs since the current state of IVR was making them worse off than if they hadn't done anything at all.

These learnings have led businesses to rethink their IVR strategies and strive for enhanced "customer experience" and "customer satisfaction" within the IVR. Businesses have adopted a multi-pronged approach to improve on the above-defined metrics (though most are far from making their IVR effective in engaging with their customers) viz.

1. Understanding customers through extensive consumer research before making any design changes in IVR menu structure
2. Measuring and tracking IVR performance through performance monitoring and reporting etc
3. Investing in advanced IVR technologies (e.g. Speech Recognition technology that recognizes spoken words) to make their IVRs more user friendly and appear "human". (Giving it a human touch has become so critical that when Citibank launched an ad campaign saying its customers can press “0” to talk to a human, it dominated the news for a while)
4. Deploying data mining and analytics to identify distinct patterns of customers’ usage behavior within IVR and utilizing those learnings for IVR menu restructuring. This trend is fast catching on and is proving to be a very effective technique in identifying customers’ preferences and issues they are facing within the IVR system.

We at Inductis (www.inductis.com) have significantly helped our clients in deploying rigorous data mining and statistical techniques for studying the call records and identifying distinct IVR usage behaviors for designing appropriate IVR strategies. I will share some of our experience in a recent engagement with a leading credit card company where we used an analytic approach to design the client's IVR strategy.

Our client was facing reduced utilization in its Interactive Voice Response (IVR) System, thus resulting in higher transaction costs as well as lower customer satisfaction. The client had started mining its IVR data as part of the effort to optimize the customer IVR experience. However, there was a gap in quickly understanding and acting upon the extensive data available.

We were asked to expand their data mining efforts and design an IVR Strategy with the following three business objectives:

1. Enhance IVR Utilization defined as % calls ending within IVR
2. Increase Customer Satisfaction defined as % calls resolved
3. Reduce Call Lengths within IVR measured in seconds

We dedicated an offshore Customer Services Analytics team to work closely with inhouse Subject Matter Experts for designing the IVR strategy and building support mechanisms.

The team worked in a phased manner. In the first phase, the team focused on diagnosing existing usage patterns. It defined important IVR performance metrics and used them to assess current IVR performance. We identified varying usage patterns across different sub-populations.

For instance, we observed people who called more frequently tended to make payments using IVR while those calling less frequently did so for changing address and not so much for making payments. Not surprisingly, we identified that older people tended to have lower IVR utilization than younger people. In addition, we identified "hot spots" within the IVR system contributing significantly to frustration and "opting-out" to an agent.

We thus established a need for statistically segmenting the caller population along their usage preferences and customizing the menu structure according to segment preferences.Our team deployed multiple statistical techniques to innovatively arrive at distinct caller segments; each with its own clear and differentiated usage pattern. The team segmented the caller population on several dimensions and established a schema which resulted in most homogenous segments capturing a large population.

In the second phase, based on the identified usage patterns and segment preferences, our Customer Services Analytics team adopted three key levers for designing strategy viz.

1. Restructure menus as per the usage preferences of segments and to address "hot spots" of frustration;
2. Add new information within menus to increase self sustainability within the IVR;
3. Remove "nuisance" value through enhanced navigation flexibility and "preferential" treatment of special cases.

Our team's redesign efforts resulted in the following enhancements within IVR which in turn would improve customer experience and customer satisfaction

1. Menus became easy for callers to use and navigate
2. Menu structure was aligned with caller usage patterns e.g. a revolver on credit card who regularly uses IVR to make payments might hear "Make Payments" as her first option
3. Once it had been made easy to navigate menus, the control of calls was effectively placed in the caller's hands, facilitating the option of opting out to an agent when needed
4. Unnecessary and irrelevant information was removed, thereby resulting in faster call resolution

Our preliminary impact analysis showed that IVR utilization would potentially increase by 4-5% while resolution of calls would likely increase by 4%. Further, we assessed that call length could be reduced by 2-5 seconds on average. All this would potentially result in savings of $4-5 Million on an annualized basis for the client. Hence the ROI for the few hundred thousand dollar, consulting investment was substantial!

This study has shown how data mining and statistical analytics can be potent cost effective tools for understanding and improving IVR in a systematic manner. Some alternative strategies suffer from one or another significant limitation viz.

1. Understanding customers through extensive consumer research is subject to sample biases and statistically invalid conclusions
2. Measuring and tracking IVR performance through performance monitoring is a reactive measure and does not allow one to identify the root causes of problems of customer dissatisfaction
3. Investing in advanced IVR technologies is quite an expensive proposition and not every business can afford it without having a robust business case for it

On the other hand, deploying data mining and analytics for understanding usage behavior patterns suffers from none of these

We predict that an increasing number of businesses will recognize the value of data analytics as the basis for enhancing their customer experience and optimizing their IVR investment. The resulting improved performance of IVR systems will clearly accrue benefits for both businesses and their customers.

• Bibliography
1. Improving IVR Self-Service; IVR Improvement Strategies 2006, The Ascent Group; www.ascentgroup.com
2. Testing your telephone based E commerce support, Journal of Electronic Commerce, Volume 12, Number 2 Is your call important? Richard Snow; www.tmcnet.com/channels/ivr/articles