Speech Technology Magazine

The 2017 State of the Speech Technology Industry: Interactive Voice Response

Becoming More Conversational
By Paul Korzeniowski - Posted Feb 17, 2017
Page1 of 2 next »
Bookmark and Share

Interactive voice response (IVR) systems have been helping businesses service customers for decades, but they still continue to gain interest. “In 2016, IVR growth experienced a resurgence and, frankly, exceeded our expectations,” says Amy Livingstone, vice president and general manager of on-premises business at Nuance Communications. “Growth was in the double digits.”

The continuing needs to create a more conversational IVR tone and to interact with customers across all channels are driving demand.

In the past few years, numerous new customer service channels, largely centering on social media, emerged, but customers still highly value voice. In fact, the phone is the most popular way to contact organizations (24 percent), according to a recent survey by Verint Systems.

Yet the phone has been showing its age, and older IVR systems have not been able to meet current needs, such as tight integration with digital channels. “We have seen a number of enterprises invest in new IVR systems as part of the refresh cycle,” Livingstone states.

IVR’s seemingly never-ending popularity piqued many suppliers’ interest. Consequently, solutions come from many vendors, including Aspect Software, AT&T, Avaya, CenturyLink, Cisco Systems, Convergys, Database Systems, Dialogic, Enghouse Systems, Five9, Genesys, IBM, inContact, Mitel Networks, Nuance, Spoken Communications, The Plum Group, Verint, Verizon, and West.

Historically, IVRs have had a spotty reputation because of the stilted way that they interact with customers. Consumers usually were greeted with robotic messages and given few options: Press 1 or Press 2. Because systems were not very user-friendly, customers spent several minutes navigating through confusing mazes of prompts and inputting mundane information, like their names and addresses. The data was sometimes misrouted or unable to move from one system to another. Consequently, many customers became frustrated and demanded to talk with live agents, increasing companies’ support costs. And then when they did transfer to a live agent, customers often had to repeat the same information they had just relayed to the automated system, adding to their frustration.

A Needed Change in Design

Newer systems are being built today to reduce input time, lower the number of reroutes, and streamline customer service. “The goal has been to make the IVR experience more personalized,” says Steven Thurlow, vice president and global practice lead for engagement management at Verint. As a result, IVR system design has been changing and becoming more intuitive, interactive, and effective.

These newer solutions go by a number of monikers, such as conversational IVR, natural language understanding, and adaptive IVR. While the terms differ, their endgame is similar: They all seek to enable customers to speak conversationally with the system and develop a reasonable expectation that the machine on the other end will understand their intent and help to service their requests.

To reach that goal, IVR systems are being buttressed with Big Data, artificial intelligence, and machine learning. Armed with those features and functions, IVR solutions support a wider range of inputs, gain a greater understanding of what the customer desires, provide more analysis of the interaction, and, most important, deliver a wider array of possible actions.

Progress is evident on a number of fronts. For instance, issues like local accents or regional dialects have historically tripped up IVR systems. Recently, though, these systems have become more adept at putting an individual’s words into context and recognizing and differentiating speech permutations.

Page1 of 2 next »
Learn more about the companies mentioned in this article in the Speech Technology Buyer's Guide:
Learn more about the companies mentioned in this article in the Vertical Markets Guide: