Analytical Views
December 2001

 

The Voice-Empowered Contact Center

BY DANIEL MILLER AND MARK PLAKIAS

What do Amtrak, ABN Amro, BellSouth Telecommunications, Delta Airlines, GMAC, OfficeDepot, Sears, T. Rowe Price, Travelocity, and Verizon all have in common? They’re all household brands with major call center operators; and they’ve all invested in automated speech recognition (ASR) systems and services in the past year. Most of their implementations fall into the tried-and-true vertical markets that have paved the way in ASR over the past few years — travel, telecom, and financial services, which have been the pioneers in this market. Increasingly, however, the interactive voice response (IVR) and ASR development teams in these verticals have been more tightly integrated with Web-based operations.

THE CHAIN, THE CHANNEL, AND THE ECONOMY
Contact centers are about as mission-critical as any IT or telecommunications resource in a business enterprise. They are at the front line of customer contact and play crucial roles in order entry and order processing. Contact centers also have “hooks” or technical interfaces into the most sensitive databases on an enterprise’s network.

Increasingly, organizations are recognizing the value of voice interfaces in the contact center. But despite recent advances, such as those mentioned above, the adoption and deployment of ASR, text-to-speech (TTS), and natural language understanding (NLU) is moving more sluggishly than technology providers would like. Clearly the slow uptake is partly attributable to the current economy, in which many IT projects are being deferred. Yet there are also other factors, including:

  • Overly-complex value chains.
  • Difficulty getting the attention of important sales and distribution channels.
  • Customer concern about flash-cuts to mission-critical applications.
  • A paucity of ready-to-go applications that plug high-volume call streams into enterprise business processes.
  • A political minefield involving defenders of entrenched call center legacies versus reluctant dragons within IT organizations capable of generating VoiceXML code.

While nobody can control the economy, the issues listed above are very real. They need to be addressed to crack the enterprise market in general and the contact center market specifically.

All too often, ASR developers have focused on overly simplified ROI models based on agent displacement. Even in instances where these displacement economics apply (and there are many), the issues listed above must be addressed before the renovation of telephone self-service can begin on a massive scale.

ASR developers and adherents need to recognize the opportunity to use voice processing technologies to enhance the user experience and stop looking at the simplistic comparison of automated speech versus live agent interaction costs. Why? Because the cost structure and work flows surrounding that agent have extended beyond a simple phone call. So, for that matter, has self-service.

HELPING THOSE WHO HELP THEMSELVES
It is important to put ASR in the context of the full scope of “self-service” technologies. These days, when contact center managers talk about self-service, they could be talking about IVR, ASR, fax, or the Web. Adding tasks for agents to perform — e-mail handling, text chat, escorted browsing, and application sharing — has impacted several core processes fundamental to any contact center including scheduling agents, reporting on their activities, delivering content, and routing contacts among agents.

These activities may have no direct impact on self-service functions, but they factor into contact center costs and metrics. Web-based multimedia contact technology has already had a disruptive effect on legacy call center core processes, including workforce management, call routing, reporting systems, and content management.

Experienced call center managers have been forced to look beyond their “pet” metrics of agent productivity, ACD capacity, and contact management in order to take into account the fact that a single “contact” can span many days and involve several channels, agents, or information resources distributed around the world.

What can we take away from all this? Voice developers seeking to recalibrate the agent workloads need to understand the total cost picture, as well as the multimedia contact technologies that are also affecting the contact center organization’s infrastructure. If an ASR application, with the help of NLU, completes a transaction without agent assistance, for example, that event needs to be fed into the connected processes of reporting, scheduling, routing, and content creation.

This is particularly challenging in the case of ASR, because it represents its own set of integration issues. Even before assessing how it plugs into recording, workforce management, and basic reporting systems, ASR implementers have to juggle a number of components within the voice recognition solution to make it all work smoothly. Fortunately, most of these pieces have existing legacy analogs that can serve as a point of departure, making the roadmap clear, and pulling along investments already in the ground.

TWO EVOLUTIONARY PATHS
We are entering the fourth decade of contact center evolution. In the beginning, there was little more to work with than ACDs handling PSTN circuits, mainframe applications, and dumb terminals, which served up text forms to agents wearing headsets. During the 30-plus years since the introduction of ACDs, they have been supplanted first by software-driven PBXs and then by communications servers with “appliances” to handle touchtone, voice, and XML-based sessions.

Minicomputers replaced mainframes, which in turn were supplanted by PCs and eventually blades that were driven by CTI middleware. Yet the most dramatic leap, at least from the end-user point of view, is taking place in the self-service continuum.

Interactive voice response units were introduced in the 1970s. They served as auto-attendants or responded to very simple queries (emulating dumb terminals). While terminals got slightly smarter or were supplanted by PCs and workstations, IVR systems persevered with little change until the introduction of “media servers” at the dawn of the new millennium. There were rudimentary ASR-based apps, rising to popularity in the mid-1990s, but the great leap forward is yet to come.

SKIP TO M’LU
This high-level view puts two scenarios for future deployment into high-relief. The first is to treat ASR as a legacy upgrade (LU) because installations will be built on the back of existing IVR platforms. This scenario will be driven by the growing demand to deliver more self-service applications over the phone, and the realization that the IVR touchtone interface can no longer support large menus and complex applications.

Some important applications fall in the LU category. Amtrak fits this description. Its Schedules & Fares touchtone application is a maze of steps, and yet represents a natural for ASR. With over five million calls a year going to its VRUs today, the conversion of Amtrak’s IVR to ASR, now underway, will expose a significant amount of consumers to the technology.

THE WEB-FACING ALTERNATIVE
The second template that comes out of the roadmap is the extension of Web-facing assets, such as the core application servers supporting self-service on the Web. Recent implementations across several of the core vertical markets — ABN Amro for Financial Services, and Lastminute.com within the travel vertical — show the basis for ASR deployment coming from Web platforms, not IVR.

In the case of ABN Amro, an example of outsourcing using service from NetByTel, a group of developers within its Mortgage.com subsidiary were instrumental in creating an ASR-based hosted solution that screened a tidal wave of calls for refinancing that emerged earlier this year as a result of declining interest rates.

By automating the eligibility screening process and finding a way to let callers know whether refinancing would (or would not) save them money on their monthly mortgage payments, ABN Amro (working with NetByTel) was able to free live agents to engage in productive inbound and outbound interactions with qualified mortgage customers. Agents freed from dealing with a surge from 7,000 inquiries per month to 30,000 inquiries were able to achieve a 60 percent conversion rate on the pre-qualified ASR prospects. That drove $100,000 in additional revenue in the first 30 days of deployment.

In the case of Lastminute.com, the largest online discount travel broker in Europe, there was no legacy IVR. The company wanted to try some cross-media development and deployed ASR as the order capture line behind an old-fashioned hard copy catalog of various gift items. Lastminute.com’s internal Web programming group worked with a voice hosting team from BT Syncordia, VUI experts from Nuance, and Periphonics voice gateway specialists to put together a solution in 12 weeks. Key elements of the platform included:

  • One hundred percent voice-automated shopping service.
  • Pick from catalogue and/or audio browse.
  • Virtual-shopping basket.
  • Credit card interface.
  • Category and product browsing interface.
  • Full postcode handling module (ZIP code=NW1 4PE).
  • Personal greeting message to attach to gift.
  • Context-sensitive help, terms and conditions, etc.
  • Confirmation by SMS, invoice by e-mail and postal mail.

What Lastminute.com’s Web team found was profoundly encouraging. While the process was not simple — a complete order consisted of 23 distinct steps — the average order, measured in terms of number of items in the shopping basket checked out, was higher than the Web. And by using a NICE logging platform in conjunction with the ASR app, Lastminute.com also found that the ability to track customer movements and intents was much more granular than the Web. This is because voice browsing and searching is much more linear, of course.

Finally, given the company’s multiple server farms spread across the North American and European communities, VoiceXML-based implementations become a natural for SIP-connected voice gateways. The company operates a distributed platform and the media resources to run a central application that connects local phone networks via the Internet.

WHAT’S THE BOTTOM LINE ON ASR?
The applications cited make it clear that a superior, automated voice user interface will improve customer satisfaction by suppressing both abandoned calls and so-called “hostile transfers” to live operators from frustrated callers struggling with complex touchtone interfaces. However legacy upgrades or extensions of Web-facing apps, ASR, TTS, and ultimately VoiceXML (or its equivalent) will converge to more elegantly blend Internet based resources, IVR systems, and live agents, the result will be improved contact center profitability, closely linked to an improved customer experience and heightened satisfaction. Automated speech can reduce lost calls, enable agents to upsell, and provide great service. As we see it, client retention and enhanced revenue opportunities — not agent replacement — will deliver strong ROI.

Daniel Miller is senior vice president and managing director, marketing and media group, for The Kelsey Group. Mark Plakias is senior vice president and managing director, communications and infrastrucure group for The Kelsey Group. For more information, visit www.kelseygroup.com

[ Return To The December 2001 Table Of Contents ]