Transforming the IVR with speech technology.

As we all know from personal experience, in the early days, there was touch-tone IVR. And let's face it--it wasn't very good. Oh, it served a purpose. As a front-end to incoming calls, at its most basic function, it could separate types of calls and least head them off into the correct general direction, eliminating the expense of hiring people to listen to long-winded caller explanations and make judgments about where to route incoming calls.

But soon, the people who configured IVRs lost the plot. I'm certainly not going to lay any blame at the feet of IVR manufacturers any more than I would blame a car manufacturer for an accident caused by a reckless driver. It wasn't the IVRs themselves, it was what call centers did with them that earned them the reputation of the second-most despised piece of call center equipment. The first, of course, being the dialer. (If we call the dialer the Typhoid Mary of the call center, we could call the old-fashioned inbound IVR the Sciatica Susan, perhaps. Third might be the Slight Cough Helen that is cheesy hold music.)

When companies began configuring their IVRs to try and take the place of humans by building their menu trees so deep and complex that more than half of users began zeroing out due to sheer, hair-pulling frustration, these companies weren't doing themselves, their customers or their agents any favors. In fact, it's hard to imagine who these companies thought they were delivering the favor to, except, perhaps, the call center manager or COO whose bonus was based on getting more people off the phone as quickly as possible. Never mind that that was being accomplished by essentially telling customers to get lost.

It's been well documented that self-service applications are notorious for raising the ire of consumers. Complaints range from not being able to easily navigate through the layers of the application to not being able to access an agent easily once it has been determined that the system can't address the consumer's issue. (Ever been kicked out of an IVR system because you didn't make the right menu choices? How long did it take your blood pressure to go down afterwards? If you're like me, it was probably weeks.) The good news is that most of these complaints can be addressed by redesigning the flow of the self-service application.

Increasingly today, the use of speech to front end self-service call routing is erasing a multitude of past sins committed by bad IVR design. Speech can not only solve many of the problems inherent in traditional IVR, it can open up a whole host of new benefits.

A well-designed, proactive self-service application can increase user adoption rates, reduce call times and create a better overall customer experience. The key to making a proactive self-service application succeed is to offer customers the right information and constantly monitor and update this information to maintain its relevance.

The Technology Behind The Curtain 

Automated speech recognition has slowly evolved from an emerging and somewhat uncertain technology to broader adoption and alternative for the push-button, dial-tone multi-frequency (DTMF) user interface for IVRs, according to industry player Intervoice. In sophisticated speech applications, it is possible to greatly reduce the steps a customer takes to accomplish a task, and also increase the caller's satisfaction in the process. Operations that track self-service efficiency are aware that customers who become frustrated with the IVR and "zero-out" to transfer to a live agent generate the most expensive calls, since call time is nearly doubled, facilities are tied up and the services of an expensive live agent are now required. When speech recognition is added to self-service operations, it is common for there to be a reduction in the number of "zero-out" transfers to live agents by customers, an increase in calls completed in self-service and a decrease in overall call handling time by the system in well-executed speech applications.

How's that for unlooked for new benefits?

"Enterprises have long sought to provide top quality customer service at an affordable price in their voice channel," said Ken Goldberg, senior vice president, corporate development and strategy for Intervoice. "Most of our customers now recognize that the use of speech enables them to offer customers an interface that is much easier to use as it allows end users to cut through long menus and improve the accuracy of self-service. In addition to reducing personnel and operational costs for the enterprise, if used appropriately, speech technology can increase customer satisfaction and retention. When a speech application is effective, it leads to very high utilization, and therefore is also efficient for the enterprise."

Trends In Voice Self-Service

According to speech-enabled IVR provider VoiceObjects, in 2007, there are two significant trends in the speech-enabled IVR marketplace as it marches forward. The first is the personalization of speech-enabled IVR systems. Personalized systems deliver customer service experiences that are automatically tailored or customized for each caller, using stored information including caller preferences, call and transaction history and so on. A recent white paper by industry analysts Daniel Hong and Ri Pierce-Grove at Datamonitor (www.datamonitor.com) declared that "IVR is ready for an extreme makeover" and recommends personalization as a critical component of the fix.

[ILLUSTRATION OMITTED]

The second broad trend in voice-enabled IVR (here's more of that "whole host of new benefits") is the increasing importance of complementary analytics and reporting technologies to answer a wide array of questions related to system usage, system and application performance, caller behavior and recognition success. Examples of these questions include:

* How many unique callers use the system? How many callers call repeatedly? How often do they call?

* Do transaction completion rates depend on age, customer status or other factors? Do different customer groups use the application differently?

* What is the average/minimum/maximum workload of the servers? Is the load on the cluster evenly balanced?

* Which prompts lead to an unusually high rate of "no input" or "no match" events? How do callers exit the application?

Try accomplishing any of that with old-fashioned touch-tone IVR systems.

With regards to what's on the horizon for the technology, Tom Chamberlain, Director of Business Process Marketing for Aspect Software, said, "One of the major trends I am seeing in self-service deployments is the increased use of a proactive application design. In a proactive self-service application, the solution automatically provides the customer with a focused set of information that will most likely meet their needs and then proceeds to offer the customer additional options. For example, a credit card customer calling the contact center can be told the balance of their card, the last payment made, the date of the next payment and the amount due at payment. This differs from traditional voice portal designs, where a customer is offered several choices immediately and is forced to navigate a pre-determined hierarchy of steps to get to the information they need."

In other words, speech-fronted IVR solutions can lead the customer through the maze rather than handing the caller a confusing map at the beginning and then forgetting about him.

How To Get Started With Voice-Enabled IVR

OK, sure, You're sold on how speech can transform your know front-end. But being clever, you know there's a little more to it than nipping down to Office Max and picking up a box of speech-enabled IVR and unfolding it in your customers' laps.

Following are some steps a company can take to implement a proactive self-service application in the most gentle manner possible, according to Aspect:

* Determine the most common inquiries in the contact center by talking to agents, listening to customers use your IVR and using quality management recording and historical reporting.

* Test the proactive self-service application on a small group of customers and collect their feedback through marketing teams.

* When choosing a customer base for roll out, pick a group that isn't technologically savvy, so that your adoption rates measurements will be more accurate.

* Communicate to your customers about what is to occur before implementation, then follow up with them afterwards.

* Over time, continue to survey customers to determine if the menu is still offering the most relevant information for them.

Slow Adoption?

Why is speech recognition not ubiquitous in all self-service operations? The answers can be reduced to two inhibiting factors, according to Intervoice, and both are being diminished by a positive convergence of industry forces. The first inhibitor to the widespread use of speech recognition is a lack of awareness that speech is now viable for enterprise operations. The second inhibiting factor is that, in the past, speech applications have sometimes been difficult and expensive to create and implement. Companies that have implemented effective speech applications have believed the benefits overcame the challenge of project complexity. These companies are realizing the reward of significant financial and operational returns on their investment.

An Example

One of Aspect's clients, a managed healthcare company, has a self-service application that uses the most popular co-pay data and proactively provides that information to members and providers when they call. This satisfies the most popular inquiries, resulting in faster service and happier customers. If a member calls and the proactive information doesn't meet their needs, they are routed to a more traditional IVR menu.

"The proactive self-service application is able to help the customers get the information they need quickly, and boost their confidence level that the data is correct," a company spokesperson said. "After implementation, we saw increased self-service adoption levels of up to 50 percent. In the future, we are hoping to implement a more natural language type of application where we can ask callers what they are looking for and by their response, we'll be able to give it to them, eliminating any menu when the person calls. The goals are to take advantage of speech, having customers use speech more and to determine why they are calling."

Imagine that. Allowing your customers to use the most natural human interface, speech, to communicate with you, instead of forcing them to navigate entering multiple digits into a two-inch square key pad in a moving car with one hand. Who ever would have thought that just speaking to the system would be an improvement?

The author may be contacted at [e-mail].

The following compaines contributed to this artice:

Aspect Sofware

www.aspect.com

Cantata Technology

www.cantata.com

Intervoice

www.intervoice.com

VoiceObjects

www.voiceobjects.com

By Tracey E. Schelmetic, Editorial Director, Customer Interaction Solutions

By Patrick Barnard, Associate Editor, Customer Interaction Solutions

RELATED ARTICLE: Speech Improves The Interaction

By Mike Chapman, Director Product Management at Cantata Technology

Speech recognition provides a more natural method of interacting with a system in a number of ways.

* Callers do not have to remember if it is "press 1" for yes or "press 2" for no; they can speak their responses.

* Callers do not have to keep pulling the telephone handset away from their ears (particularly important when, for example, using a cell phone in the car) to interact with the key pad.

* Speech recognition can obscure the differences between IVR systems: Any one person may use several different IVRs on a daily basis (office, cell phone voice mail, home, etc.) and the need to remember different menu choices for the same function, deleting a call, for example, is eliminated. The caller need only say "delete."

* Using speech allows service providers to support different IVR systems without having to retrain their users to new systems with different menu choices.

* Speech recognition can obscure the differences between handsets: most wireless and cordless phones have the # and * keys in a similar positions, whereas most PDAs do not.

* Speech recognition eliminates the cumbersome search for the # and * keys.