Curing the "zero-out" habit while improving overall customer satisfaction.

For many companies, the thought of introducing speech automation to customers immediately conjures up an unpleasant compromise: saving the company money while risking customer dissatisfaction, and a zero-out to speak with a live agent. Not only is there the chance of negating the savings, there is also a chance of endangering goodwill with the customer and losing his or her business.

**********

Personal experience plays a strong role in molding opinions about automated self-care options. Many of us have encountered bad Web sites and horrendous touch-tone or "press-or-say" systems that actually encourage customers to build a "zero-out to an operator" habit. Recent advancements of technology and standards in the speech recognition area, however, are having a remarkable impact on engaging callers in more friendly and flexible dialogs without human intervention. A portion of the improvement in customer satisfaction with speech automation can be attributed to an evolution of speech standards. Standards within speech are creating a foundation for increased personalization and the ability to engage callers in exchanges that more closely mimic a conversation with a live agent through the support of recent advancements in database access and call flows.

The Trickle-Down Effect Of Standards

The primary benefit of standards within a technology is not always apparent to the end user. According to Webster's, a "standard" establishes a generally accepted model "by authority, custom or general consent." One of the most obvious benefits of technology solutions for end users is investment protection, whereas a lack of standards can lock someone into a relationship with a specific vendor with a proprietary solution. Callers trying to access information from a contact center do not care about these issues; they simply want their questions answered as quickly and efficiently as possible.

Evolving standards in the speech market are directly affecting how callers interact with automation and are providing investment protection for those deploying speech systems. One of the many benefits of the VoiceXML (VXML) standard within speech is the ability to deploy a distributed architecture that closely resembles Web-based models. One result is that it makes it easier for contact centers to repurpose existing data for speech, allowing detail such as account balances and personal information to be injected into audio dialogs that were previously the exclusive realm of a live agent. The ultimate benefit for callers is that more pertinent information can be combined with even better speech recognition and user interface design. An automated solution becomes a pleasant experience, boosting customer satisfaction.

[ILLUSTRATION OMITTED]

Why Should Your Callers Care About VoiceXML?

Actually, callers don't care about VoiceXML. What they care about are ease of use, performance, reliability, access, etc. In many of these areas, VoiceXML contributes to increased value for contact centers that choose to explore not only how to reduce costs but also how to improve the quality of each customer interaction. It's not that VoiceXML is a secret sauce for success so much as it is a standard and supporting architecture that enables speech solution designers to more creatively craft applications that exceed expectations.

VoiceXML is the presentation logic for most speech systems being deployed today. A VoiceXML browser is the primary component used to retrieve and parse application code, as well as to handle data exchange between speech recognition/TTS servers and database access. Definition of the VoiceXML standard within specific boundaries allows vendors to create highly distributed architectures.

For example, the VoiceXML browser, speech recognition software, media server both for prompts (i.e., audio guidance) and for grammar (i.e., words that can be recognized), application server and caching server could all be in separate boxes and even distributed across multiple locations. This might initially seem unnecessarily complex. However, the value beyond performance considerations is flexibility of ownership, control and security. One result is integration of speech beyond what was previously considered a stovepipe application within the telephony organization.

This is where the rubber meets the road for callers with respect to VoiceXML architecture. Vendors are incorporating existing data for callers to enable consistency across multiple contact channels, such as Web-based and even agent-assisted interactions. Security also plays a role, as contact centers can now take advantage of the hosted speech systems while maintaining sensitive information within their facilities.

A great example of this approach is that of a large financial services firm. Customers using their telephones can check their balances, move funds or even find the nearest ATM. All sensitive information is maintained within the confines of the bank, while critical telephony and speech hardware and software required to answer calls and navigate automated dialogs are hosted by an outsourced speech service provider. Results do not only include improved customer service, as company executives are also better able to focus funds and internal personnel on core banking infrastructure. This new hybrid architecture is enabled in part by the flexibility of VoiceXML.

Application Server Is the Key

A Web-based model for speech is actually a good analogy. An application server within a VoiceXML environment operates very much like a Web server for creating pages for viewing over the Internet. A speech application server maintains the VoiceXML application with its associated call flow, prompts, grammars and direction for database access. Uniform resource locators (URL) are used to identify where the appropriate data or prompt files must be retrieved.

This model frees the speech application from what was previously a difficult-to-update static system. Pre-VoiceXML systems usually locked companies into a hardware model wherein the system administrator had to be educated on myriad proprietary procedures and coding schemes. An interesting byproduct of the architectural flexibility of VoiceXML is the ability to control the application server ("the brain") by the company or a partner offering a packaged speech solution.

Performance Is Critical

Performance is a critical issue for speech systems, especially when distributed systems introduced wide area networking connectivity and in essence more moving parts that could fail. Dead air is one of the worst of all outcomes in a speech system, as callers simply do not have any of the visual cues available with Web-based access. As a result, systems invisible to callers but essential to success must be in place. These include caching servers to supply grammars for speech recognition, geographic redundancy for all servers and telephony access, as well as constant monitoring for all components, from database access to local/wide area network uptime and server uptime.

Security

The opportunity to take advantage of a distributed architecture enabled by VoiceXML can reduce the risk of a security breach by deploying a distributed architecture. In this scenario, the data repository for a speech system is independently located and secured from the rest of the speech system, whether it is in a hosted environment or simply located elsewhere within an organization. However, emphasis must be placed on the security of the links by utilizing virtual private networks, mutual authentication, etc.

It's easy to pigeonhole VoiceXML as just another protocol in the never-ending alphabet soup of technology jargon. The difference with this standard lies in that it is encouraging new methods of delivering automated contact center solutions that are having a positive impact on customer satisfaction.

The catalyst is not simply a maturing speech market where consistent code promises application portability. It is also the embodiment of a Web-based architecture that is gaining traction across many industry segments in both large and small organizations. The results are that many new speech applications are shattering previously conceived notions, with automated solutions that dramatically reduce the chances that a customer will zero out while greatly improving overall customer satisfaction.

For information and subscriptions, visit www.TMCnet.com or call 203-852-6800.

By Ted Cwiok

Convergys Corporation

Theodore (Ted) Cwiok is general manager of the Speech Solutions Group of Customer Management Group (CMG) for Convergys Corporation (www.convergys.com). Prior to his current position, Cwiok served as vice president of Corporate Technology for Convergys' CMG, where he was responsible for corporatewide technology, including applications, computer operations, telecommunications, end-user computing and wide area networks. Previously, Cwiok served in a similar role at MATRIXX Marketing, a Convergys predecessor. He also occupied the position of vice president of Development and Operations at Cincinnati Bell Information Systems (CBIS), also a Convergys predecessor.