Though total market penetration remains relatively low—at 28 percent, according to research firm Frost & Sullivan—contact centers continue to be the primary customers for speech analytics solutions. Most of those contact centers that have already installed speech analytics solutions did so to improve their operations. “Most were about improving efficiencies, like first call resolution, handling time, and moving calls to self-help, all of which were derived from the need to drive costs down,” says Barak Eilam, president of business applications at NICE Systems, a speech analytics systems vendor.
But because there is a limit to the costs that can be eliminated, the greatest benefit of a speech analytics deployment comes when company officials are allowed to apply solutions more broadly across the entire organization, not just toward improving call center operations.
“Although it has significant value to the contact center organization, limiting it to the contact center is not using it to its full extent,” Eilam says.
By leveraging speech analytics, companies can proactively identify sales opportunities and uncover customer satisfaction issues—both of which will greatly benefit the sales and marketing units. Other divisions, including product management, research and development, human resources, manufacturing, billing, compliance, collections, fraud prevention, risk management, and accounting, can also derive tremendous advantage from the value-generating business intelligence uncovered by an effective speech analytics solution.
Yet for all the information speech analytics can provide, very few organizations understand how to fully exploit the data, or the technology itself, as a strategic business tool.
Many organizations are further constrained by limited resources and siloed operations. Worse still, few have a firm grip on who should own—that is, be responsible for—the speech analytics application. Marketing? Sales? The contact center? Accounting? The IT staff? Corporate leadership?
Application ownership implies responsibility not only for the initial purchase and rollout, but also for all of the follow-up work that is involved long after a solution goes live. Many people live and work under the false assumption that speech analytics is a form of plug-and-play technology; in reality the applications require a significant amount of work and resources to make them function properly. Users constantly have to fine-tune applications, improve definitions, realign parameters, and then work and rework facets of the applications until they get them right for their specific operating environments. Once that is achieved, everyone has to understand that if they’re going to use speech analytics effectively, then they’re going to need to dedicate resources to it.
That’s why establishing overall ownership of an application is one of the hardest parts of any speech analytics deployment, according to Elizabeth Herrell, vice president and principal analyst at Forrester Research. “There are so many names, titles, and roles of people who can use speech analytics, but who wants to do [what’s necessary] is the key going forward,” she says.
“I’m not sure there is one clear answer, or one best answer,” Eilam admits. “It’s not one size fits all. It’s a lot about how an organization is structured internally.”
Daniel Ziv, vice president of customer interaction analytics at Verint Systems, agrees. “It depends what you plan to do strategically with an application,” he says, “and there’s a lot you can do with analytics today.”
Adding to the chaos is confusion within most organizations about who actually owns the customer, and that’s not something that can be clearly defined either, Herrell says.
Often, the marketing department thinks the customer experience falls entirely within its domain, but it’s not that simple. “If you think of analytics as a sophisticated reporting tool for the customer experience, it could be any number of people or departments, depending on the type of organization it is and the service or product it offers,” Herrell states.
For that reason, she thinks it’s better to divide application ownership based on the types of calls received. “If it’s a marketing problem, it belongs to marketing. If it’s a sales issue, it belongs to sales. If it’s a billing issue, it belongs to accounting. It could also be a service delivery issue, and that goes to product management,” she says. “It’s all about where the problem resides, and that’s why we have IVRs—to get the caller to the right person or department to help with his problem or issue.”
But because speech analytics can touch so many areas within the corporate structure, is splitting responsibility for an application a wise idea? Not at all, suggests Donna Fluss, president of DMG Consulting. “Enterprises need to have one central repository,” she argues. “It doesn’t matter who oversees the centralized function, but it has to be centralized.”
At some companies, that responsibility falls to one person, but those instances are rare. “I get really impressed when I see a business card with a speech analytics manager/business analyst’s title,” Fluss says.
Such a title might exist at a handful of companies—probably less than 10 percent, according to Herrell. The more likely scenario, she says, is “many executives within a company who are responsible for disseminating the information to many other executives.”
The ideal scenario, many experts agree, is for the application to fall under the auspices of a business team with the political clout to share its findings across departmental lines on a timely basis and to help departments take action to realize identified opportunities. In a handful of companies, that body is a Six Sigma team—comprised of experts specially trained to evaluate existing business processes and determine ways to improve them, to remove defects and inefficiencies, or to design brand new processes. Six Sigma teams, however, are common only in the larger contact centers with thousands of agent positions. “Once you get to 100 to 200 seats, companies do not have the same overhead to put teams like this together,” Ziv says.
Still, many others argue that speech analytics is, by and large, a call center technology, and responsibility for it should, therefore, reside in the call center. “It’s most effective use is in the contact center, and to that end, the contact center people should own it,” says Keith Dawson, principal analyst at Frost & Sullivan. “They’re the ones that will use it and operate it.”
Dawson further contends that the contact center is the one piece of the corporate puzzle that can put all of the information gleaned from a speech analytics solution into context. “The contact center has to be the funnel for delivering the information to the rest of the company,” he says. “To say that a customer was dissatisfied, the information is not actionable unless [all parties] know the context, like how long the customer waited in the queue, whether the agent response was appropriate, or if the IVR misdirected him. That’s why the contact center needs to have the application under its belt: so it can provide an organizationwide understanding of what’s going on with the customer.”
Fluss also thinks the contact center should have ultimate responsibility over the speech analytics solutions. After all, calls originate at the contact center, and the recordings are made at the contact center, she says.
The contact center is constantly hammered with calls from upset, less-than-satisfied consumers, and the contact center is the only place within the corporate structure “big enough and equipped to handle them all,” she adds.
Early on, contact centers were the central figure in speech analytics rollouts, and Dawson thinks that’s still a good idea for companies just starting out with the technology. “The call center has to be the first place where you take a stab at putting it in. You’re probably not going to get it into a company unless the call center is driving the application at first,” he states. “It has to be about boosting [agent] performance first and the customer second. The customer interaction is not the main driver yet, but it will be in about three to five years.”
That’s why the early deployments often saw the head of quality monitoring/training inside the call center taking ownership of the application and establishing the metrics.
Herrell, on the other hand, points out that the contact center no longer holds a monopoly on customer interactions. “We’re way beyond the stage where the contact center owns the customer,” she says. “It can’t be limited to the first point of contact any more.”
Instead, Herrell and several others see speech analytics reaching a point where the contact center manager does not have be directly involved. “Reports can be generated automatically, and you’ll need experts to detect the flags that require action,” she says.
“A growing trend we see now is for people from marketing exposed to these applications,” NICE’s Eilam adds. “Marketing is starting to take ownership. People there are more data-savvy and tech-savvy. And the business analysts are working for the marketing department and taking responsibility for all the data, not just marketing data, but billing data, sales data, etc.”
But despite this lack of agreement over who should ultimately have control of a speech analytics application, the one universally accepted principle is that it should not rest solely in the hands of the IT department. “It’s not an IT tool but a business tool,” Ziv explains. “Nothing [about speech analytics] should be a pure IT decision.”
Eilam agrees. “IT’s role is shrinking,” he says. “We see in some cases where IT is trying to take it over, but [speech analytics] is not just about technical skills. More and more it’s about business skills.”
That’s not to say that the IT department should be excluded entirely. It should have a role in purchasing decisions, especially given the fact that it will be responsible for supporting and maintaining the system, Forrester’s Herrell points out. “It’s often a good idea to have someone within the IT department who is responsible for evaluating products and software helping to determine what they can and should get,” she says. “IT should be responsible for the technical aspects [of an application] rather than doing the deep dives into what affects customers.”
Dawson doesn’t mince words when it comes to IT’s role. “I would never want to say that IT shouldn’t be involved because they definitely should be,” he says, “but IT can also act as a barrier. Organizations need IT flexibility to make the cross-departmental aspects work, but IT continues to work with applications as siloes, and they need to stop that.”
And that’s not necessarily unique to the IT department. “If [the speech analytics application] is being driven just by IT, marketing, sales, etc., it’s going to be a siloed business tool. It needs to be driven by the contact center’s need to be more productive,” Dawson continues. “If it doesn’t start in the contact center, it will fall flat. It will not be as effective as it could be, and given how much applications cost, you definitely want to maximize their effectiveness.”
That’s a lot of pressure for any company, and for those that don’t want any part of speech analytics application ownership for that reason, a number of technology and equipment providers are now packaging their offerings via hosted, managed services models. Though adoption of hosted speech analytics has been slow thus far, these types of offerings are getting a boost from firms looking for more flexible payment models that limit capital expenditures while keeping them current with technology changes. They are also appealing to companies that want an objective, unbiased, and external source to provide customer insight, according to Verint’s Ziv.
But even more than that, they are appealing to companies of all sizes for very different reasons. For small to midsized businesses, a hosted service is a surefire way to get the benefits of a solution without having to carry the full financial burden of capital expenditure or maintenance costs or having to allocate personnel or other resources that are probably not available in the first place. “But we’re also seeing traction among very large enterprises because they do not want to coordinate the efforts between the many corporate divisions involved,” says Eilam, whose firm launched a fully managed, hosted version of its Interaction Analytics business solution in mid-June.
Telrex, another speech analytics vendor, launched CallRex Speech Analytics as a pay-as-you-go service that also includes a consulting service whereby consultants work with the business to determine what is driving customer behavior, to monitor calls for compliance, to identify missed sales opportunities, to improve customer satisfaction scores, and to leverage call recordings as a business asset.
“By offering speech analytics as a service, we are lowering the cost of entry and making speech analytics technology accessible to businesses of every size,” says Robert Kapela, president of Telrex.
An added benefit of any hosted solution is that it allows call center managers to monitor agents on the phone regardless of their geographic location.
Through the hosted model, a third-party firm, like Verint, NICE, or Telrex, can take and record the calls, process the data, and draft reports for the client as directed. The vendor can deliver the reports at predetermined intervals, such as daily, weekly, or monthly, or post the data to a secure, password-protected server or Web site that companies can access on their own.
Some companies prefer to maintain their own recordings and have someone else do the deep dives into their contents. Others prefer to have nothing in house, but that can be a dangerous proposition, Ziv says.
“You may not want to outsource your customers fully,” he states, noting that the last thing a company wants to do is become fully dependent on a third party to determine its customer strategy. “They should take some ownership of their customer experience,” he says.
That mode of thinking is one of the key factors in the low adoption rate thus far, according to Frost & Sullivan’s Dawson. “We’ve not detected a groundswell of interest [in hosted speech analytics] just yet, partly because of security concerns, partly because of bandwidth concerns, partly because of the complicated nature of the applications, and because they are tied into so many other things related to the business,” he says.
All of these concerns “are not particularly unsolvable, but the industry has not really looked at them yet,” Dawson adds.
The Big Package
Vendors have also been adding their speech analytics solutions as part of a much larger quality monitoring, quality assurance, workforce management, or business intelligence suite. In fact, several analysts have noted that very few companies are filing RFPs for workforce optimization solutions without having some form of speech analytics as one of the main requirements. Customers and prospects have expressed interest, but it has been limited to those that could afford to buy in.
In the past year, however, a number of smaller vendors have entered the market with solutions that are pared down, so they have less functionality, but they also come with a lower price tag. For many, at least this gives them a way to get into speech analytics for the first time without breaking the bank.
This sales model often involves “creating speech analytics solutions that are designed to be less expensive and less complicated,” Fluss explains. “It’s intended to get customers up and running with speech analytics relatively quickly.”
But the world itself is getting more technologically complicated, and one need not look too far beyond the contact center for proof. Few can argue that an increasing number of customer contacts are now taking place via fax, text messaging, email, Web chat, customer surveys, and other nonvoice interactions. This brings even more confusion to the question of who should own the analytics applications that pull them all together. After all, who owns the Web site? Where does the fax machine sit?
But many argue that it shouldn’t matter. It’s still all about the customer interaction, regardless of how it’s coming in, Verint’s Ziv says, “and it’s the same issues affecting the people who handle those interactions.”
“It still should all be perceived as a contact,” NICE’s Eilam adds.
As the world becomes more multimodal, analytics solutions will need to take on a more broad-based approach to integrate structured and unstructured data from many channels and sources. In tandem, end users are likely to expect their tools to focus more on the analysis and less on the speech. And that could be a mixed blessing for an already downtrodden speech analytics industry.
Taking SA Out of the Call Center
While most people identify speech analytics as a call center technology, modern-day solutions typically got their start decades ago in the military and government sector. The intelligence and law-enforcement communities first used speech analytics to record, monitor, search, mine, and analyze intercepted conversations, broadcasts, and other communications.
Today, speech analytics is finding uses in areas far outside of the government and contact center environments. One of the latest is in voice message management, which involves using speech analytics technologies to identify the content of a recorded voice message, categorize the message, assign it a priority, and determine to whom it should be delivered.
“Anywhere that you have lots of conversations that can be legally recorded, speech analytics has great potential and would be practical,” says Donna Fluss, president of DMG Consulting.
Speech analytics software is also an area of great interest to search engine companies like Google and Yahoo! So it’s no surprise that search, particularly voice search, “is another great potential application,” Fluss says.
Perhaps nowhere is that more relevant today than in the myriad of speech that is locked in video. Speech analytics can provide much-needed access to the audio information locked within the video itself.
“With more and more video and audio clips on the Internet, there’s a lot more data out there,” notes Daniel Ziv, vice president of customer interaction analytics at Verint Systems. By using speech analytics to sift through that content, “companies can get an idea of what people are talking about and what they really care about,” he adds.
The same applies to mining television and radio broadcast streams, for example. Elizabeth Herrell, vice president and principal analyst at Forrester Research, calls this one of the greatest growth opportunities for speech analytics. “You could look through newscasts to see how often a person or event is mentioned,” she says. “Really, it can be applied anywhere that there is a large repository of spoken words.”
Another of those areas is the legal arena, where speech analytics is now allowing litigators to quickly search hundreds, if not thousands, of audio recordings that in the past would have taken many hours to do. In addition, companies in financial services can closely monitor their businesses across a much larger sample of conversations for regulatory compliance.
Speech analytics should not be considered just a standalone application. It is also highly useful as a module that can bring value to users of CRM applications, e-learning applications, and business intelligence applications.
“There is a large, relatively untapped market for interaction recording systems in smaller call centers and beyond them, to related businesses like branches and back offices,” says Keith Dawson, principal analyst at Frost & Sullivan. “These organizations haven’t in the past been able to justify the costs of a full-fledged, compliance-grade recording system, especially since so few of them have on-site access to IT resources.”
Still, despite all of the potential ways in which speech analytics could be used, both in the present and the future, “they all have issues that will have to be worked out over time,” Fluss says.