Speech in Need of a Bailout?

Article Featured Image

It’s not clear how many of the banks, brokerage houses, credit unions, and lending institutions that received federal bailout money earlier this year used some of those funds to improve their automated customer self-service systems, but that is one area that definitely could use a little improvement, according to a 2008 financial services industry report card of sorts from Vocal Laboratories (VocaLabs).

The report cited widespread “room for improving both the caller satisfaction levels and the operational efficiency of call centers.”

This comes despite the fact that financial services firms long ago started aggressively pursuing speech technologies to improve customer service, accommodate a growing mobile user base, and cut costs. In fact, financial services firms have often been at the forefront of speech technology deployments, installing new technologies long before most other industries.

“Banks are desperate to get people out of the human-to-human contact scenarios,” says James VanDyke, president and founder of Javelin Strategy & Research, a financial services industry research firm.  The costs of these human-to-human contacts, whether they take place at the branch office, over the phone, or via email—banks still need a live person to read and respond to email—are “just killing banks today,” VanDyke says.

“Sure, banks have been early adopters of the technology, but very few actual [fully automated] transactions are happening as a result,” he continues. “There’s so much more that could be happening as compared to what’s actually happening.”

According to Javelin’s research, 14 percent of all U.S. households have sent an email to a bank representative within the past 90 days, but perhaps more telling, 10 percent of all U.S. households spoke to a live person over the phone within the past seven days, 16 percent within the past eight to 30 days, and 41 percent within the past 90 days. By most estimates, as many as 70 percent of all inbound calls to financial services firms during any given time period are from customers seeking basic information, like an account balance, stock quote, or the location and hours of branches and ATMs, all of which could be related fairly easily by automated systems.

“That’s just an awful lot of people who could be using speech technologies,” VanDyke says.

Further compounding matters, in their efforts to scale back the human-to-human element, many financial firms have ended up trying to force automation on customers, even when it might not suit their needs.

“The net result is frustrated callers, multiple calls to accomplish a single task, general discontent with the experience, and higher operational costs than would be the case with better automation,” VocaLabs said in its report. “These systems are generating a significant number of complaints, and should be redesigned to ensure that they are as effective and enticing for callers as possible. The key is improving the quality of the automation to the point where callers prefer it to a live agent.”

Among the problems cited by those who took part in the VocaLabs survey were:

  • difficulty reaching a live agent;
  • difficulty using, understanding, or navigating automated systems,
  • long hold times;
  • poor agent skills;
  • low call completion rates, often because the system led a caller down the wrong path, forcing her to hang up and dial again; and
  • long average call handling times.

The current economic climate is not helping either. Financial services firms are seeing an increase in the number of calls from investors, account holders, and customers worried about their assets.

“It’s not business as usual. Certain events are spiking up the number of phone calls, like questions about a 401(k) or moving money around,” says Daniel Ziv, vice president of customer interaction analytics at Verint Witness Actionable Solutions.

Also contributing to this spike in call volume is a high number of requests to renegotiate and modify the terms of existing loans and mortgages. According to Hope Now, a Washington-based coalition of mortgage lenders and counselors, more than 1.6 million mortgages have been modified since 2007.

“Mortgage companies and financial services institutions are faced with managing this incredible volume of requests. They require tools that are specifically designed to help them manage the call volume while still driving a superior customer experience,” says Andrea Ayers, president of customer management at Convergys, which last month released just such a solution.

The Convergys Loan Modification Solutions line leverages the company’s live agent assistance, on-demand self-service automation, analytics, and proactive outbound communications via voice, email, or text. The solutions are available as a full suite or as individual services.

Better in Bulk?

Call volumes are not just rising on the inbound side. As hard luck falls on many consumers, lenders and credit card companies have had to step up their collection efforts, and that has so far meant an increase in the number of outbound calls as well. These calls can be expensive, so to justify the cost lenders have also had to improve the collection rates they get on those calls. Traditionally, though, more than half of the agent’s time could be spent simply trying to get through to the right person and then authenticating him.

But despite the industry’s shortcomings with voice and IVR technologies and their modern usage, financial services firms continue to hunt for new ways to incorporate voice automation and self-service into their customer-facing operations. Among them are voice-based commerce, with receipts sent by text message to mobile phones; one-click call support from within an online account; outbound voice alerts regarding account activity, status, and transaction history; and promotional voice messages, according to voice platform provider Ribbit.

Other activities seen as great candidates for automation with speech are ATM and credit card activations, collecting loan application information, paying bills by phone, confirming that a check cleared, and customer surveys.

Additionally, enterprises, especially in the financial services sector, “continue to invest in workforce optimization technologies, customer relationship management, and speech analytics,” says Daniel Hong, lead analyst at Datamonitor.

The reasons for this, especially when it comes to speech analytics, are simple. “Pushing customers to self-service is one thing, but when something goes wrong, customers will call in, and you’ll need to do the math to fix what’s wrong with self-service,” Verint’s Ziv says. “In every call center there is always low-hanging fruit that can be addressed, but you need to know what the key drivers are.”

Ziv also notes that not only are current fiscal trends driving more financial services companies to look at analytics, but those that have already been using the technology are now looking to expand existing applications. “People have seen speech analytics as very strategic, but now they’re looking for it to be more operational,” he says. “They’re looking for something more proactive, to identify trends that they may not have been aware of and what is contributing to the calls coming in. They can use analytics without having a team of analysts to pore over the data.”

This trend is being accelerated by strong growth in the use of call recording, with many firms now recording 100 percent of their calls and holding onto the recordings for as long as a year. A prime contributor to this is a need for banks, brokerages, and investment firms to protect themselves,and their assets during conflict resolution.

But even under the best circumstances, and with the most advanced call recording, analytics, speech recognition, and other technologies in place, customers are still uncomfortable providing personal details over the phone to call center representatives, according to the recently published 2009 Salmat VeCommerce Identity Verification Study. The study points to a growing consumer fear that traditional PINs and passwords do not adequately protect their personal information, with more than 50 percent of respondents expressing the belief that someone could guess their PINs, passwords, or other security details, and 59 percent believing that someone else already knows those details. In addition, 37 percent of respondents had either experienced identity fraud or theft themselves or know a friend or family member who has fallen victim to these crimes.

Financial services firms, perhaps more than most others, have the most to gain by tightening security on customer interaction channels. The Salmat VeCommerce study found the most preferred method of verifying identity was biometric voice identification. This was favored by 45 percent of respondents, followed by PIN (21 percent), password (18 percent), and personal details or history questions (16 percent).

But because no one method is 100 percent foolproof, many financial services firms are now turning to multifactor authentication, using a combination of technologies and methodologies for added protection. Voice authentication—which verifies the caller’s voice model against one that is on file—is often being combined with traditional user ID, password, and personal information requests during phone interactions. On site at the branch office, authentication methods that can be used include biometric verification, such as fingerprint scanning, iris recognition, facial recognition, and voice ID, in addition to smart cards and other electronic devices.

VanDyke sees the benefits of voice authentication going far beyond the call center, though. “The area of authentication is very significant for voice recognition,” he says. “You can give an individual total control over his own finances. After the bank authenticates him, he can be talking directly to his account, asking it what its balance is, sending money to his son in college, transferring money between accounts, etc.”

Closely tied into this is the use of speech technologies for password resets. Not surprisingly, many people have problems remembering the myriad passwords they use every day to access everything from email and voicemail to their cell phones and bank accounts. Also not surprising is the fact that so many calls to internal help desks and customer service centers are requests for password resets.

Voice solutions can alleviate this resource-intensive process via a self-service channel that uses the power of speech recognition and voice biometrics. By simply answering a series of personal questions or reciting a unique identifier, a user can automatically reset his unique password, which can then be disclosed over the phone or issued via email.

Here, too, it’s all about cost reductions. If financial firms can do away with having people at a help desk fielding password reset requests, then that cuts a huge piece of overhead, Javelin’s VanDyke says.

Even more important, cutting these kinds of human-to-human contacts  frees staff to dedicate more time to revenue-generating activities, such as dispensing investment advice or closing a mortgage.

“These are things that are not as easy as they were last year when everything was a lot more predictable,” Verint’s Ziv says.

And no firm, big or small, is immune. “Because of the times, everyone is looking for better ways to cut costs and do things better,” Ziv states, noting that economic conditions are even forcing small and midsize firms in the financial services sector to pursue speech technologies. Speech use, he says, “has been higher in larger call centers, and SMBs have been watching what their big brothers have been doing. Now they’re ready to go ahead and make the investment too.”

Little wonder, then, that vendors are gearing solutions to smaller firms, and making them more affordable by offering them in a hosted environment. And, unlike in the past when financial services firms shied away from hosting for security reasons, new and tighter methods of encryption are taking some of those worries away. “There’s a lot of encryption for information storage and retrieval. There are a lot of ways you can do this now,” Ziv says.

Speech and the ATM

Talking money machines might not be in the cards just yet.

Since the first self-service interactive kiosk was deployed at the University of Illinois at Urbana-Champaign in 1977, customer acceptance of the technology has exploded.
In fact, NextGen Research estimates that by 2013, more than 2.6 million financial kiosks and ATMs will be in use around the world. But at least in the United States, where more than half of these kiosks and ATMs are expected to be deployed, don’t look for speech output as a standard just yet.

The two largest ATM manufacturers, Long Beach, Miss.-based Triton Systems and North Canton, Ohio-based Diebold, offer the technology on most machines they produce, but U.S. banks and credit unions have not been rushing to deploy it.

For years, it was feared that banks would have no choice. Dating back to 2002, advocates on behalf of the disabled began pushing the financial services industry to make speech output a natural part of ATM transactions.

But while pressure continues to mount for voice-enabled ATMs in the U.S., the truth of the matter is that banks are not required to offer it, and, therefore, most have chosen not to.

“With regard to implementing speech, there’s nothing now,” says Robin Springer, president of ComputerTalk, a consulting firm specializing in the design and implementation of speech recognition and other hands-free technologies. “There aren’t any mandates to say that they have to put in speech. Banks aren’t required to put in voice, just to make [ATMs] accessible. Most banks in the U.S. are just putting Braille characters on the machines.”

But while U.S. financial services firms have been reluctant to implement voice in ATMs, many banks throughout Canada, Europe, Asia, and Australia have embraced the technology. Canadian Imperial Bank of Commerce, for example, installed embedded text-to-speech at its approximately 3,800 ATMs, allowing its visually impaired customers to plug in special headsets and hear what appears on the ATM screen. “Bank machines allow our customers the ability to hear private voice instructions on how to conduct an [ATM] transaction with audio headphones,” the company explains on its Web site.

A number of French banks, including CIC, Crédit Mutuel, BNP Paribas, and Societe General, have all within the past few years also included speech synthesis technology in their ATMs to assist blind and visually impaired customers. A year ago, CIC and Crédit Mutuel began incorporating Acapela Group’s text-to-speech technology into their ATMs so that when a client plugs a headset into a special socket on the ATM, the machine switches from showing information on screen to a vocal assistance service. With it, the client is guided in either French or English on how to insert his bank card, enter and confirm the amount to be withdrawn, enter his PIN code, and retrieve his money, receipt, and bank card. Around the same time, Diebold, through a partnership with Nuance Communications, began installing ATMs equipped with Nuance’s RealSpeak software at BNP Paribas’ branches.

And though Springer thinks that all ATMs will be voice-enabled one day, right now she argues that U.S. banks don’t have an incentive to explore the technology, which can cost between $1,000 and $3,000 per ATM. “What’s the benefit to me in putting in speech recognition or text-to-speech? Right now it’s so much cheaper and easier to go with Braille,” she says. “And there are all sorts of issues with headsets, like cleanliness, safety, sanitation, and theft.”

Perhaps the day will come when banks begin to realize they can use voice in ATMs for more than just servicing their visually impaired customers. Banks can also use voice technology over ATMs to personalize transactions, place advertisements, and read statements.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues