When combining self-service and traditional call center components, it is important to clearly define the call-flow process so that all interactions work seamlessly. The solution should incorporate built-in contingency plans that connect customers with a live agent as soon as additional assistance is required. Savvy companies achieve the best of both worlds by enhancing the customer experience while reducing the overall cost of customer care. Lowering the unit cost of customer interaction is a complex process that involves weighing the pros and cons of numerous interconnected variables.

In the 1970s and 1980s, most speech technologies concentrated on routing calls through IVR systems, handling simple inquiries, and generally yielding very low customer satisfaction results. Over the past three years, advancements in conversational technologies have enabled the deployment of sophisticated and engaging personas to handle more complex transactions with an improved customer experience. Conversational access applications have moved from the research lab to the IT center as a valued solution.

The next step is to develop and refine processes, methods, and technologies in speech applications. Voice is the most natural and standard interface for communications. Voice applications represent a powerful medium in a global economy for improving the dynamics in channeling care.

In 2003, with the broad acceptance of VoiceXML the industry saw a major shift in the way conversational applications were deployed. Business logic became separated from the user interface, creating an environment where back-end processes could be leveraged and reused across multiple modalities. VoiceXML dramatically reduces the time and cost of deploying new conversational solutions, and IBM has augmented those benefits by creating graphical tools based on the open standard Eclipse framework. Using these advanced VoiceXML generating tools, developers can visually map out the call flow of their applications.

Speech has gone through all of the phases of a startup as it moves from a lab experiment, to a few good deployments, to a mature technology that can have a significant impact on the call center, generating greater customer satisfaction, customer retention, and additional revenue streams. Teams in IBM Research have made tremendous progress in natural language understanding (NLU), the study of spoken word recognition and unstructured dialogue interactions. The most common use of NLU today is for call-routing systems that use the opening dialog "How can I help you?" and then use NLU to parse the users' utterances and transfer them to the appropriate person or conversational application. More sophisticated NLU applications, such as the one deployed at T. Rowe Price, allow you to complete a broad set of transactions, such as buying or selling mutual funds, through a conversational interface.

Users expect to be able to customize their experience when they are dealing with a conversational interface with multiple applications. Call center managers want to be able to add and delete services without having to rewrite their entire applications. As both audiences begin to embrace new technologies in speech, we will witness the transition from analysis to implementation, resulting in optimum solutions for driving down costs, while shifting investments to the strategic dimensions of the care program.