Featured Research
Financial institutions can reap significant financial gains by using voiceprints to authenticate through the phone channel. By our calculations, a large money center bank will see over $100 million in economic benefit from reducing fraud loss in the contact center, lowering operating costs, and providing a more pleasing customer experience.
This report is made available courtesy of Victrio. Click here to view the report summary. Contact Pete Headrick (pheadrick@opusresearch.net) to receive a copy
For more information on becoming an Opus Research client, please contact Pete Headrick (pheadrick@opusresearch.net).
IBM launches Watson Engagement Advisor; Entry Level Virtual Assistant for “Smarter Commerce”
Two years after Watson made its high profile debut on TV’s Jeopardy, IBM is introducing a cloud-based making its “cognitive computing” resource available as a cloud-based service to support customer care. Big Blue introduced Watson Engagement Advisor at the SmarterCommerce Global Summit in Nashville, TN. At the same time, spokespeople noted that an impressive roster of companies were already gaining experience with the service through an early customer program (ECP). A partial list of participants includes RBC (Royal Bank of Canada), ANZ Bank (Australia), USAA, Verizon, Telstra, Allstate Insurance and Celcom Axiata Bhd (Kuala Lumpur).
The announcement includes an engagement model of its own. IBM expects companies to be able to have their Watson Engagement Advisor up and running within 6 weeks and that enterprise customers who put it to use can expect to reach their ROI goals within 6 months. That six week period is used to identify specific content and “Q&A Pairs” to give Watson a starting point in its cognitive journey. This information is uploaded electronically to a version of Watson in the cloud and becomes the basis for building initial “custom UIs” to handle the most commonly asked questions. The original information is augmented by accompanying documentation, such as product manuals or sales collateral in PDF format or information from the company’s Web pages (HTML).
When the information is loaded, a pilot begins among a subset of employees or customers. During the pilot, Watson is able to expand its vocabulary and build what amounts to a larger “grammar” of terms or questions that it understands. At the end of the 6 week period, IBM’s professional services personnel migrates the pilot into production mode, where use by the broader population will further improve its ability to understand and answer questions. Once it demonstrates that it can scale in production, enterprise implementers can begin to train the system on other domains or sets of “Q&A,” with the help of subject matter experts among company employees or customers.
The ECP started in March and has already shown results which IBM is showcasing at the SmarterCommerce Global Summit. It is showcasing the ability for a computer system to assist in customer engagement by enabling them to converse in “plain-English” dialogs. It uses past conversations and feedback to refine future answers, thus it improves over time. IBM’s value proposition is that this approach will deepen relationships with customers, leading to higher retention rates and (theoretically) enhanced revenue. At the very least, much like FAQs, automated handling of the questions that arise most often will reduce operating expenses associated with live reps in contact centers.
The new packaging and promotion by IBM feels like a simple way for enterprises to get started and get comfortable with virtual assistants. From this modest beginning we can see how enterprises can expand upon their offerings, making them a cornerstone to personalization and, eventially, morphing them from “assistants” to “advisors.” As a large, global provider of enterprise infrastructure software and middleware, IBM is in a great position to raise the profile of Personal Virtual Assistants (PVA). The IBM Watson imprimatur (though it can be confused with AT&T’s Watson platform) has good name recognition as well.
[Updated and corrected] The roll-up of mid-market, cloud-based customer engagement resources is accelerating. Today the Boards of Directors at both Genesys and SoundBite approved a definitive agreement for Genesys to buy all of SoundBite’s shares for $5.00 each (a total exceeding $100 million). The purchase price is a hefty premium over the $2.99/share price at the close of business on Friday and reflects the high expectations on the part of management for Genesys to move up the latter among companies in the $1.3 billion “Conversational Cloud” market space.
Genesys has been engaged in a long march toward an expanded cloud offering for some time now. It had already launched a home-grown Software-as-a-Service (SaaS) offering, Genesys Cloud Connect For Salesforce, when it made the decision to acquire cloud-based customer-care provider Angel from parent company MicroStrategy for a reported $110 million in February of this year. All told Genesys’ cloud-based businesses, including acquisitions, generate roughly $135 million. Given the growth rates in hosted customer care that figure could easily eclipse $150 million in the coming calendar year. That would approach the estimated revenue for category leader West Interactive (which had acquired Holly and Tuvox), and it would compare favorably to other top providers: Convergys (Interactive Voice Platform) and [24]7 (which acquired both Tellme and Voxify).
The SoundBite acquisition augments Angel’s cloud-based clients and those supported by partners. It will bring Genesys’ total number of of cloud-based end-customers to 1,250 directly supported, plus uncounted clients whose applications are supported by hosting partners. The latter are those end-customers that are served by incumbent telco’s like Orange Business (France Telecom), Verizon Business, AT&T, BellCanada and CenturyLink (formerly Qwest). In other instances, Genesys may find itself selling against long-time go-to-market partners like EchoPass and Teletech. I almost forgot that a trifecta of long-time “pure plays” in the Voice-in-the-Cloud Universe collectively account for 25,000 ports worth of the Genesys Voice Portal (GVP) operating in their own clouds. They are Nuance on Demand, Voxify (as noted, now part of [24]7) and West.
SoundBite, now a public company, will be taken private by the Genesys acquisition. The last full year’s accounting of financial performance is represented by the 10-K form it filed with the United States Security and Exchange Commission (SEC) last March. In it, one learns that the firm’s top line revenue grew 10% per year between 2010 and 2012 (from $39.9 million to $48 million in 2012). However, during the lean years from 2008 to 2010, the revenues declined an average of 4.9% each year (from $43 million to $39.9). During that time, SoundBite never showed a profit. Its highest level of annual loss was $4 million in 2009, which has been trimmed to $1.8 million in 2012.
Prior to the this transaction, SoundBite had been conducting a roll-up of its own in the mobile marketing arena. Acquisitions and diversification fueled SoundBite’s revenue growth in 2011-2012. It acquired SmartReply in June 2011, which gave it the ability to add email and text messaging to its portfolio of “Proactive Customer Communications,” which were largely voice based, outbound notifications. In February 2012, the company acquired mobile marketing specialist 2ergo Americas (the U.S. based operations of the UK-based company). Collectively the companies added about $2.7 million each to SoundBite’s top line. More importantly, they augmented the hosted voice services that were largely targeted to large contact center customers to include SMS, mobile coupons, QR code, and mobile web platforms.
Today, all these assets, platforms and customer lists belong to Genesys.
In response to demand for contact center resources that integrate well with e-commerce Web sites, Genesys Labs formally introduced a new product called Proactive Engagement at its G-Force customer care forum in Boca Raton. The product is an amalgam (or recombination) of proven components of existing Genesys software, including Genesys Composer, Genesys Administrator, Genesys Statistics Server and Conversation Manager, along with deep integration to the core Genesys routing engine to make sure that conversations are delivered to the best agent in a timely manner, or what Genesys refers to as “The Moment of Truth” (MoT).
Proactive Engagement is able to recognize the Moment of Truth by monitoring an individual customer’s past activity on the company’s e-commerce Web site, including page views, the entry of text into a dialogue boxes, mouse overs and the like. These events are incorporated into a unified customer record, where they could also be merged with records in the CRM database, as well as other metadata, as part of separate integration activities. The focus of Proactive Engagement is to make sure the initial chat session and subsequent call routing is informed by past customer activity, as well as the availability of the best agent for the task.
As an example, an electronics retailer will be aware that a customer has visited the online showroom to click on the description of a few HD TVs. He may even have loaded one or more into a shopping cart, which may have been abandoned. The next time the customer visits the company Web site, Genesys Proactive Engagement can display an invitation to chat with an agent who is prepared to answer specific questions and help make a purchase decision. If no agent is available, other methods to contact the customer can be invoked, including email or other marketing initiatives.
The software to support Web monitoring, notifications and interconnection with Genesys routing and other back-end processors is licensed as a single platform for a one-time fee. Other elements, such as the plug ins for the Interaction Workspace (Genesys’ agent desktop) and links to the chat server and statistical reporting are licensed on a “per seat” basis. The company reports that 17 customers have already joined its “early adopters” program and are proving the value of using Proactive Engagement to increase revenue opportunities and provide better one-to-one service once a customer is connected to a live agent via chat or over the phone.
McAfee’s LiveSafe Suite uses Face Recognition and Voice Biometrics to Protect Personal Information
Security software specialist McAfee (an Intel Company) has introduced a comprehensive new service called LiveSafe, to help individuals protect “unlimited number of PCs, Macs, smartphones or tablets.” Core capabilities start with basic protection from malware or viruses on Windows, Android and iOS-based devices. It also addresses some of the new problems that a mobile lifestyle entails, including an “anti-theft” function that helps “lock and locate” lost devices, as well as a password management feature based on McAfee SafeKey, which provides a single password for accessing the many username/password combinations that plague online denizens.
A novel feature of LiveSafe McAfee is its ability to secure a customer’s “digital assets” using two biometric attributes, both voiceprint and facial recognition are used to provide fast, secure access to up to 1 gigabyte of encrypted data. McAfee expects LiveSafe users to upload such things as financial records, passports or important legal documents, such as deeds or trusts so that they can be accessed from all of the supported devices.
LiveSafe is a first-of-its-kind combination of personal information management, identity verification resources and security services. It reflects the fact that the folks at McAfee understand how living the mobile life and being constantly online through a variety of devices exposes people to constant security threats. Banks and telephone companies have tried to convince folks to upload important documents for protection and today almost a dozen firms around the world have developed elements of the personal data ecosystem.
It makes sense for individuals to look to a a security-oriented technology provider for such a service, especially when they have gone the distance of adding biometric authentication which, by definition, is the only authentication factor that is “something you are.”
LiveSafe is being offered at a promotional $19.99 annual subscription. After the introductory period, the regular price will be $79.99.
Voice Biometrics Conference 2013-San Francisco came to a close yesterday after a succession of presentations and panels that highlighted the growing need for better authentication and the role that voice biometrics can play far beyond its original presence alongside IVRs (interactive voice response systems) in enterprise contact centers. We started with a keynote presentation by Michael Barrett, chief information security officer for PayPal and (perhaps more importantly) the chairman of the FIDO Alliance. He described how and why the clunky and largely ineffective use of UserName and Passwords for personal authentication is ready for its end-of-life.
He highlighted how protocols, platforms and products that conform to the architecture endorsed by the FIDO Alliance (which stands for “Fast ID Online”) will enable end-users (ideally, all people) to select the means of authentication that they find most convenient at the time they need to assert their respective identities. We started with such a broad discussion – “Let’s Replace Passwords” – to play up the fact that the world is ready for alternatives to current ways of authenticating and that using one’s voice as an identifier will have real value in an increasingly mobile, hands-free and eyes-front world in which biometrics are being taken seriously, and no single factor will suffice.
Participating by executives from Google, Nok Nok Labs, Validity and Agnitio, in addition to PayPal, in a subsequent discussion demonstrated how seriously the initiative is being taken by industry influencers and creative start-ups, alike. The protocol, along with several working use cases, is already very mature. It reflects a world where individual credentials and authentication mechanisms move to the access devices and are directly under the control of the end user. This should have great appeal to device makers and their owners, and it already started discussions at the conference surrounding whether banks, retailers or other businesses with financial stakes will find authentication at the endpoint to be acceptable to them.
To highlight the global nature of the voice-based authentication phenomenon, the next enabled the manager of contact centers and alternative channels from Avea, a large GSM/wireless provider in Turkey to show how his company registered the voiceprints from over 1.5 million subscribers in order to replace challenge questions, which both customers and contact center agents were finding to be annoying and expensive. The implementation uses software and systems from Turkish speech processing specialist Sestek and provided proof positive that solutions can scale quickly and provide rapid payback by shaving more than 20 seconds from customer care calls.
We followed with a presentation from a different technology alliance – VoiceVault, Geo Semiconductor and SpeechFX – to show how the speech recognition, speaker recognition and gesture analysis can be integrated into home electronics to support highly-personalized interactions with the family television. Then Day One came to a close with yet another partnership describing how enterprises are meeting the security challenges that are exacerbated by mobile workers, multi-channel commerce and high levels of real time collaboration. The speakers were from IBM and Nuance and reflected how voice biometrics, in conjunction with other authentication methods, help companies deal support secure communications in a “multi-perimeter world.”
Day Two kicked off with another validation point for voice biometrics. As multiple newspapers around the world were happy to report, Barclays Wealth and Investment advisors are using passive enrollment and passive authentication to simplify caller authentication to the point where customers can securely carry out high value interactions. As a sample, in the UK, the Telegraph’s reporter observed that “Barcleys Wealth customers will no longer need to answer security questions and remember pin codes to use telephone banking.” Similar stories appeared in the Wall Street Journal, ZDNet, Gizmodo and several technology and business outlets, with quotes from VBC-SF’s keynote speaker Matt Smallman, who is in charge of Customer Experience, Strategy and Change at Barclays.
Smallman’s talk was punctuated by a roundtable of executives from a variety of financial institutions, including Bank of the West, Vanguard, and Union Bank. In sum, they described how changes in customer usage patterns, especially online banking and mobile applications necessitate close evaluation of alternatives. At this point, only a handful of banks have integrated biometric-based authentication, but they are well-aware of the strengths and weaknesses of the alternatives before them.
Beyond banking, Day Two highlighted real-world use cases for voice biometrics, starting with a presentation from NICE Systems to describe how to “operationalize” the technology by leveraging existing contact center resources, especially call recording and analytics systems. This was followed by a thought-provoking discussion of “What You Can Learn from a Phone Call,” featuring the likes of Pindrop Security, TRUSTID and Voxeo, which are all companies that have developed technologies that can determine the risk associated with a phone call and its originator based on unique data about such things as the originating number, line characteristics, “device signatures” and the like.
In addition to deep technology discussions, Day Two featured a number of “reality checks” giving solutions providers a chance to describe the state of the technologies and general acceptance of such things as using “black lists” and “white lists” for fraud prevention (via Victrio), adding face recognition (SpeechPro), mobile enrollment and authentication (Agnitio), eGovernment implementations (VoiceTrust) and embedded solutions for “Truly Handsfree” authentication and speech recognition (Sensory). The intent – which attendees tell us was largely successful – was to provide solid basis for assessment of “market maturity” which is always dependent on whether the underlying technologies perform as advertised (which is largely a “yes”) and whether businesses and device makers are ready to put them to use, which is largely dependent on whether they see demand from their own customers, clientele or citizens.
We closed with a panel of executives to talk about where we are and where we’re going. The consensus is that no single technology or architecture is about to dominate in the world of strong, convenient authentication. Therefore, solutions providers must be prepared to support all channels and form factors. I, myself, think that future success will depend largely on marketing efforts of service providers, device makers, retailers and even automobile manufacturers. It starts with linking strong authentication with “personalization,” while making spoken words or commands a preferred method for authentication. Though the largest implementations have, thus far, been in contact centers and associated with IVR platforms, the most promising areas for growth are taking place at the edges. That means in the background (supporting passive enrollment and authentication) and at the other extreme, inside individual devices or components.
Once that sort of positioning is accomplished, you’ll see the potential for voice-based authentication to ride the coat-tails of the personal virtual assistants, automotive entertainment systems and the next generation of remote controls for home entertainment and utility systems.
Nuance Communications has issued two press releases that emphasize the game changing role that voice-based authentication is playing to improve customer care. The releases coincide with Day One of Voice Biometrics Conference – San Francisco. The first release confirms that Barclay’s Wealth & Investment Management unit is employing the FreeSpeech (text independent) voice biometric platform to support “passive” enrollment and authentication of clients. Simply stated, Barclay’s can gain strong confidence in the identity of a caller by monitoring the conversation between a client and his or her financial adviser and comparing it to a voiceprint that has been generated from past conversations. Clients are not required to enroll by repeating phrases, nor do they have to say that phrase in order to carry out their business.
The second release provides results from two recent surveys that show how this sort of passive authentication appeals to the public at large. When the results of a recent survey conducted by Nuance of 900 smartphone users are combined with those generated by Coleman Parkes and Opus Research in a report issued last July, you get a picture of a frustrated set of consumers ready to take advantage of new technologies that reduce the time it takes to carry out their business both over the phone and online.
Customer frustration results from failure to remember PINs, passwords or a set of challenge questions (many of which are not of the caller’s making). One of the major themes at Voice Biometrics Conference – San Francisco is that passwords can be rendered passe. We have the technologies and we have the demand. Barclay’s approach is one example, but there are many more. For instance, the conference agenda leads off with a keynote from PayPal’s Michael Barrett, who is also the chairperson of the FIDO Alliance (Fast ID Online Alliance), a consortium that is poised to replace passwords with a number of multi-factor alternatives.
A brave, new world of highly-personalized and secure interactions over mobile phones and electronic devices in cars and homes once the process of authenticating is made trivial. The Barclay’s example will be followed by many more and initiatives like the FIDO Alliance are just getting started.
GEICO, the diversified insurance company owned by Berkshire Hathaway, has teamed with Nuance Communications to offer its own mobile, personal virtual assistant (PVA) branded as Lily. The company hasn’t made the sort of marketing splash that it made when it introduced its mobile app with a media blitz starring Maxwell the Pig yelling “wee wee wee” in the backseat of a family station wagon, but that is because management feels like it doesn’t have to; if they do it well, mobile customers will naturally use it.
In a press release issued when the app was updated in the Apple App Store, GEICO cited the fact that, even prior to the formal introduction of a conversational element to the service, 40% (“two out of five”) of the people with the app “use voice technology on their mobile devices today.”
GEICO’s experience is not an aberration. A survey conducted by OnlineDegress.com about a year ago showed that nearly 9 out of ten iPhone 4s owners used Siri at least once a month. More than one out of four used it daily to check email. Only 35% said that they would never use Siri. In short, they were on a mission of discovery and, for something like 40% of users had found the Siri-enabled functions that they were ready to use regularly. More importantly, more than half of the respondents said that having a Siri-like feature would be “critical” in terms of choosing their next smartphone.
Speech Skeptics are bound to express doubts about wanting to talk to a machine while seeking roadside assistance. Their hesitation is based on bad past experience with earlier speech recognition systems, perhaps in noisy environments. Critics also tend to treat “virtual assistants” as the ugly stepchildren of speech enabled “personae” – meaning machines programmed to have a certain “attitude” that many users found grating or annoying.
Indeed, the claim from Pete Meoli, GEICO’s director of mobile and digital design, that Lily would have “a lively personality to allow our mobile customers to connect with her at a deeper level” will give some prospective users a cause for pause. But that’s not the point. Lily – as the natural interface to the GEICO mobile app – will have the ability to support interactions at a “deeper level.” She’ll be location aware, she’ll have account history at her fingertips and, on top of all that, she’ll benefit form fairly accurate speech recognition and a corpus of candidates for “next best action” distilled from voluminous amounts of GEICO’s historical interactions.
As often as I’ve called my automobile insurance provider and spent the obligatory 5 minutes making selections from a series of voice menus, or sitting on hold waiting to talk to the one person who specializes in my particularly problem, I know that interacting with an cool, calm, collected automated assistant would be a welcome, positive change. It just takes some getting used to.
Featured Research
The market for voice biometrics-based solutions has matured significantly in the past year. In this document Opus Research compares the products and strategic position of a select group of solution providers and presents structured analysis that should benefit decision makers planning to introduce speaker verification in their contact centers or on mobile devices or networks.
Featured Research Reports are available to registered users only.
Click Here to View the Report Summary
For more information on becoming an Opus Research client, please contact Pete Headrick (pheadrick@opusresearch.net).
Back in December, we noted that an SF-based start-up called Expect Labs was working with Nuance Communications (and others) to deliver a showcase iPad application called MindMeld. The demo was an impressive display of what the company calls “Anticipatory Computing.” Like Google Now, it is the product of constant monitoring and analysis of human input (words, gestures, location…) for the purpose of delivering relevant items (search results) from a set of defined sources on the Internet or World Wide Web. In the case of MindMeld, for example, if two people start talking about “happy hour” near the end of work day, the app may display specials and offer suggested watering holes based on local information from Yelp! or other local search sources.
That’s the low-hanging fruit. It was enough to garner a reported $2.4 million in seed financing from Google, Greylock, Bessemer, IDG Ventures, KPG Ventures and Quest Venture Partners. This week, Expect Labs added three new investors whose contributions “more than equaled” the previous investment. More importantly, it signals great future expectations in three distinct opportunity areas. Device manufacturer Samsung has never been shy about adding new features and functions that define the “natural user interface.” With its competitive sights set squarely on defeating Apple’s iOS-based devices in the marketplace, Samsung is motivated to continually differentiate. Spoken words, gestures, touch and other signals of user intent must be taken into account. Expect Labs’ core technologies will be expected to make the most of matching all that input with relevant responses.
For Intel, the value of what Expect Labs should be equally obvious. Making sense out of voluminous amounts of diverse inputs will be extremely computing-intensive. Expect Labs is creating products and services that showcase the power of what Intel calls Perceptual Computing, which is Intel’s term for technologies designed to render the mouse unnecessary by replacing it with more natural user interfaces, including speech, gestures and touch. Intel has issued a software development kit (SDK) to encourage creative input from app developers around the world. All the different modalities of input represent more grist or other raw material for Expect Labs’ core technology platform, the Anticipation Engine.
The participation of Telefonica Digital represents a wise investment by a very forward-looking, global telephone company. Parent company, Telefonica, is the seventh diversified telecommunications company and the fifth largeset wireless carrier. Telefonica Digital was formed last year as a global business unit focused growing a diverse set of digital assets through R&D, acquisition or partnerships. It has identified cloud computing, mobile advertising, M2M and eHealth as focus areas, but management well understands that fostering and delivering applications and services that leverage “Big Data” and “Cloud Computing” to serve over 315 million customers will be its growth engine. Expect Labs’ Anticipation Engine, residing in Telefonica Digital’s cloud and hoovering up user input and scads of real-time data suits Telefonica’s requirements very nicely.
There is a lot of fertile ground where Perceptual Computing meets Anticipatory Engines. The resulting products and services will define one of the families of Virtual Personal Assistants (VPAs) that are destined to redefine mobile search and ecommerce. Results from the Anticipation Engine already have many similarities to the latest features of Google Now, which uses a form of Anticipatory Computing to provide relevant information to support daily planning (like weather, travel plans, traffic conditions and other factors like restaurant reservations or movie show times). Solutions developers are making great strides in figuring out when, where and how it is reasonable to expect this type of automated system to be of assistance to individuals. It is a long-term learning process by both humans and machines.