September 2004

By Roxann Swanson, Nortel Networks

Welcome to the new and improved world of speech automated self-service. Once a 'nice concept, but not reality' application, the adoption of speech recognition within the enterprise is poised for breakthrough in the coming years. In the Gartner report 'Speech Recognition Telephony Software: Worldwide 2002-2007,' analysts said the speech recognition market is poised to reach more than $258 million by 2007, with contact center and business portal applications the largest segments for adoption.

Why now? In a nutshell, it comes down to customer loyalty and the need to keep customers happy. Customers will readily adopt such services, with the following caveats: they must be quick, simple and reliable. Additional important by-products of speech recognition and contact center tools include improved employee productivity, lower operational costs and increased revenue. But ultimately speech automated self-service solutions are about the customer.

Customers are more savvy and smarter than ever, with more technology choices before them. Every interaction customers have with your enterprise must say, 'We value your business, we value your time and, most important, we value you.' Enterprises must provide consistent quality communications at each touchpoint. This creates two very important complementary results: You minimize customer frustration, while also providing this precious constituent with the right information quickly.

But this begs the next question ' how? How can an enterprise implement or upgrade current customer contact applications and improve the engagements it has with its customers? And how can an enterprise do so without spending large sums of capital?

Step One: Identify the business transaction.
You must think about what type of transaction you want to enable and how to do so. Transactions well suited for speech automation include simple, repetitive transactions such as account balancing or order status reporting, as well as more complex transactions such as obtaining a stock quote for one of 15,000 companies listed on a stock exchange. Regardless of the complexity of the transaction, companies need to find an easy way for customers to interact with them, as this presents an ideal opportunity to improve customer loyalty. For example, some banks permit customers to carry out financial transactions at a time suitable to them by simply speaking their instructions. The instructions are converted into commands that are executed on the bank's mainframe computers. Responses are provided to customers in natural speech. The customer does not need to sort through a confusing multitude of options; nor does the customer need to be on the phone for an extended period. By understanding what type of information a customer will seek, companies are better able to ensure their customers get the information they need quickly.

While this is the primary benefit of the interactive voice response (IVR)-enabled approach, another is reduced costs for the organization in two key areas: time and resources. Rather than present a long list of alternatives, which requires a customer to stay on the phone longer, a speech-enabled IVR helps the customer cut to the chase. A call that usually takes 90 seconds can be reduced to 30 seconds. Also, if a customer can receive the information quickly through the IVR-enabled approach, there's less chance he or she will engage a company representative. This creates a more effective deployment and use of call center representatives, freeing them up for more critical customer service needs.

Step Two: Evaluate current investments.
You must evaluate the type of infrastructure you currently have in place. Then you must determine if your contact center is relatively up-to-date (5 to 7 years old). Also, you need to consider your plan to derive the maximum value out of your legacy systems and, in particular, you must determine how you can extend or enhance your contact center with new technologies. A converged voice and data networking solution that blends IP telephony-enabled applications with legacy systems protects an existing investment while also enhancing the system. It enables seamless transfer of voice and data information associated with that call to an agent's desktop, so the caller does not have to repeat information.

IVR platforms and Voice Extensible Markup Language (XML) will work within an existing infrastructure to manage multiple forms of communications securely and reliably. The VoiceXML standard allows enterprises to leverage back-end Web infrastructure and enables companies to provide access to the same information and transactions via the phone as on the Internet. For example, a customer can transfer money on the Web and then use the phone to hear the updated balance. Other features to consider are end-to-end project management, training, installation and ongoing maintenance and enhancement services.

Step Three: Pick the right tools.
To meet your specific business requirements, you need to create and deploy flexible advanced speech recognition applications based on various speech recognition tools (large vocabulary recognition, natural language understanding, speaker verification and text-to-speech). This will require a mix of both standardized and customized applications, as each plays an important role in a comprehensive speech recognition deployment.

An application portfolio that consists of a large number of standardized applications with completed design, coding and testing helps ensure you have a good foundation on which to build your solution. From there, you can customize applications with specific prompts or host interfaces that map to your specific business needs.
In the end, you must make the customer your primary focus. In this context, speech recognition applications are becoming a vital element in a top-notch CRM environment. With the proper system integration, reliability, scalability and investment protection, you can make these technologies a vital component of your organization's competitive advantage. Done right, a speech recognition system can yield shorter calls, increased accuracy of information provided, elimination of dreaded long hold times and, in the end, a more satisfied customer. By-products include reduced agent expenses and well-coordinated transaction processing. Done wrong, you have one very unhappy customer. In this economic environment, that's the one thing no one can afford.

Roxann Swanson is vice president and general manager, enterprise communication applications for Nortel Networks (www.nortelnetworks.com)