FEATURE The conclusion of all this may be that if recognition technology can continue to improve, and deliver not just face but full body and motion recognition, then the resulting ‘win:win’ combination of superior systems and super recognisers should mean law enforcement and border control are able to identify and catch more criminals.
References 1. ‘Super-recognizers: People with extraordinary face recognition ability’. Richard Russell, Brad Duchaine, Ken Nakayama. Psychonomic Bulletin & Review, Volume 16, Issue 2, pp 252-257. April 2009. 2. Katrin Bennhold. ‘London Police ‘Super Recognizer’ Walks Beat With a Facebook of the Mind’. New York Times, 9 October 2015. Accessed March 2016. http://www.nytimes.com/2015/10/10/ world/europe/london-police-super-recog-
nizer-walks-beat-with-a-facebook-of-themind.html?_r=1 3. Melanie Hall. ‘German police bring in British “super-recogniser” cops to identify New Year sex attackers in Cologne’. The Telegraph. 24 January 2016. Accessed March 2016. http://www.telegraph. co.uk/news/worldnews/europe/germany/12117655/German-police-bring-inBritish-super-recogniser-cops-to-identifyNew-Year-sex-attackers-in-Cologne.html 4. Robert Ratcliff. ‘London Riots: Only 1 Arrest Made as Result of Facial Recognition’. IFSEC Global. Accessed March 2016. http://www.ifsecglobal.com/ london-riots-only-1-arrest-made-as-resultof-facial-recognition/ 5. Kirsty McMahon. ‘Law enforcement divided over effectiveness of facial recognition’. Security News Desk. 3 November 2014. Accessed March 2016. http://www. securitynewsdesk.com/law-enforcement-
divided-effectiveness-facial-recognition/ 6. Kevin Rawlinson. ‘Facial recognition technology: How well does it work?’. BBC News. 3 February 2015. Accessed March 2016. http://www.bbc.co.uk/news/technology-31112604 7. ‘Unaware Person Recognition from the Body when Face Identification Fails’, A. Rice, P. J. Phillips, V. Natu, X. An and A. J. O’Toole. Psychological Science, 24 (11), 2235-2243. 2013. ‘Recognizing people from dynamic and stable faces and bodies: Dissecting identity with a fusion approach’. A. J. O’Toole, P. J. Phillips, S. Weimer, D. A. Roark, J. Ayyad, R. Barwick, J. Dunlop. Vision Research, 51, pp 74-83. 2011. 8. P. Jonathon Phillips. ‘Improving Face Recognition Technology’. Computing Now. March 2011. Accessed March 2016. https://www.computer.org/web/computingnow/0411/whatsnew/computer
Contact centres: how to escape the games of ’Guess Who?’
Amit Desai
Amit Desai, Verint Systems Can you confirm your postcode? What’s your mother’s maiden name? What are the first, fourth and fifty-fourth letters of your secret password? In the face of such an onslaught of security questions – the logical consequence of passwordbased authentication – customers are forced to rack their brains for the right answers and it is easy for frustrations to mount. Consumers are getting increasingly annoyed about this style of authentication process, which has become more like a long game of ‘Guess Who?’ than a customer service interaction. The danger of failing to authenticate customers properly is highlighted by the rising levels of identity theft. According to TransUnion, 19 people fall victim to fraud every minute1. Targeting anything from banking to mobile phone accounts, thieves are becoming more sophisticated when it comes to stealing identities and turning a profit. The wider impact is substantial; recently Financial Fraud Action UK2 found that the cost of telephone banking fraud rose 95% to £14.4m in the first half of 2015, compared to the same period the year before – and the contact centre remains a key focal point of vulnerability. 8
Biometric Technology Today
However, the contact centre is also a vital engine of customer satisfaction and loyalty. When it comes to providing a positive and memorable experience, consumers value speed and convenience above everything else. A recent study of customer service that Verint commissioned in association with Ovum3 found that 81% of consumers just want their questions answered and on average, almost half feel that those companies that deal with their requests quickly deliver a better experience than those that don’t. There’s no question that businesses need to work harder to combat the potentially crippling
impact of fraud, without compromising on the customer experience; not only do such cases damage business reputation, companies also face hefty fines if found in breach of customer data and security regulations. But in some ways executives are being asked to strike an impossible balance – between engendering customer loyalty on the one hand, and guaranteeing the security of their data on the other. So in the effort to fight fraud and accurately identify customers, what’s the alternative to ‘Guess Who?’ style questioning? The answer is: definitely not more security measures. Instead, organisations need to make better use of existing touchpoints to make security smarter.
Dual-ing against fraud Contact centres are failing to take advantage of significant advances in voice biometrics that
April 2016
FEATURE have drastically increased the accuracy of analysing and identifying voice signatures in real time, while also slashing call times and saving businesses tens of millions in eliminated fraud losses. The new generation of voice biometrics operates ‘passively’ and recognises a caller’s voice without requiring them to speak a passphrase. In tandem with this growing technological sophistication, the introduction of a ‘dual’ approach to voice authentication (where two almost identical technologies work in tandem to keep customers happy while also fighting fraud) has further helped to streamline and strengthen these systems. Firstly, the customer authentication application holds ‘voiceprints’ for each customer, analysing the current caller against these previous signatures. Most of the time, this will allow the customer to be authenticated quickly, reducing the frustrating experience of having to reel off reams of facts about their past. The accuracy of this method of authentication is high – comparing one voiceprint to a database of similar voiceprints is straightforward. Sophisticated systems will continue to build on their original voiceprints for each customer, creating an increasingly diverse identifier that allows quick authentication when the right customer calls.
“Contact centres are failing to take advantage of significant advances in voice biometrics that have drastically increased the accuracy of analysing and identifying voice signatures in real time, while also slashing call times and saving businesses tens of millions in eliminated fraud losses” The second string of the approach includes a fraud detection application that stores a database of voices of known fraudsters, to identify criminals on the phone before they’ve had a chance to act, even if they are able to provide the right password and answer all the security questions. Granted, this system compares one voiceprint to potentially thousands of known criminals, so the chances of catching a criminal in the act aren’t as high as matching a customer to their voiceprint. But as the database grows, so do the chances of catching the felon red-handed.
Beyond the passphrase A key advantage of ‘passive’ voice biometrics is that it is a frictionless approach, match-
April 2016
For contact centres, one key advantage of ‘passive’ voice biometrics is that it is a frictionless approach, matching a caller’s voice to their stored voiceprint without them having to speak a passphrase.
ing a caller’s voice to their stored voiceprint without asking them to speak a passphrase. At worst, this application doesn’t locate any matches and the contact centre can revert to the traditional security question; nothing is lost and the customer continues the verification process. But more often than not, the contact centre is in possession of a useful mechanism, able to catch known scammers in the act and help stop their customers becoming victims of fraud. With these two voice ID processes layered on top of one another, a balance is struck between streamlining the customer experience – by authenticating each customer before they’re subjected to a barrage of security questions – as well as ensuring security. The technologies are also virtually ‘unspoofable’; as the voice is essentially a vocal fingerprint, it’s incredibly difficult for a fraudster to fake a customer voice well enough to fool the authentication tools. Of course, there are factors that make legacy authentication systems less successful (such as asking the user to say a certain phrase as a passcode), which make these systems more prone to customer pushback and frustration. Not only this, but if the customer’s voice doesn’t match the database due to external factors such as a bad line, the contact centre has simply added a further step to an already frustrating identification process. Fortunately, as the technology has evolved, so have these dated systems. Newergeneration platforms are replacing the need for any pass phrase, in turn making the technology more likely to be adopted by consumers.
Banks leading the charge Several industries – particularly those that hold sensitive customer information, such as the banking sector – are widely adopting fraud detection applications. For banks, the stakes are high if fraudsters succeed in penetrating and taking over customer accounts, so having
the strongest capabilities in place helps them to uphold reputations and save potentially millions in fraud losses. There’s also evidence to suggest customers are more trusting of organisations, particularly in the banking sector, that are taking clear steps to ensure their customer data is kept safe. While our research with Ovum suggests that one in five consumers don’t trust any businesses to keep their information safe, 43% do trust their bank with their data. This highlights the long road ahead for brands trying to build consumer confidence, but it also shows that banks are on the right path to cementing this relationship with their customers. Compare this with sectors that have been much slower to adopt the technologies, such as the utilities industry; a miniscule 3% of consumers trust their utilities provider to keep their data safe. It’s evident that those investing in security measures are benefitting from greater customer trust and loyalty.
Security over service Having said that, there are a number of objections that mean the customer authentication area has seen a much slower uptake of newer voice biometric systems. Again, some of the larger banks are beginning to use these technologies, but this pales in comparison to the uptake of fraud detection technologies – and the reason for this is partially down to business priorities. KPMG predicted that fraud cost businesses in the UK £732m last year4, demonstrating the obvious financial implications of a fraudulent incident. Beyond that, attacks can also significantly damage brand reputations, and in turn levels of consumer trust.
“Some of the biggest business concerns over voice biometrics relate to the perceived feasibility of obtaining a ‘perfect’ voiceprint. But these objections are based on a fundamental misconception of how the technology works today” Data breaches like TalkTalk’s highlight the impact that a fraudulent cyber attack can have on consumer trust and loyalty5, so businesses are keen to avoid this above all else. Consequently investment tends to focus on avoiding such a disastrous attack, as opposed to improving the customer experience. For
Biometric Technology Today
9
FEATURE tion. Voice biometric technology picks up much more minute nuances in the individual’s throat, larynx, mouth and nose, to form a digital voiceprint. So while these voiceprints are liable to variations – much like our written signatures – the underlying characteristics are less prone to change. It is these underlying features that voice biometrics is designed to detect. The resulting voiceprints are therefore truly unique, even in identical twins. As a result, today’s voice biometric systems can reliably authenticate callers by phone irrespective of their circumstances.
Big Brother society Voice biometric technology picks up minute nuances in the individual’s throat, larynx, mouth and nose to form a digital voiceprint. The resulting voiceprints are therefore truly unique, even in identical twins.
customer authentication, then, while the business case of cutting down contact times and improving customer experience is desirable, security is a more essential business priority. Beyond this, some of the biggest business concerns over voice biometrics relate to the perceived feasibility of obtaining a ‘perfect’ voiceprint. Background noise, illness and even ageing all contribute to changes in our
10
Biometric Technology Today
voice. So the difficulty of getting a perfect voiceprint is often seen by contact centres as another barrier to solving the customer query; if you are unable to authenticate the customer’s voice then you’re slowing down the customer experience. But these objections are based on a fundamental misconception of how the technology works today. When we as human beings hear a voice, we hear accent, tone, words and emo-
A more fundamental problem is that the privacy issues surrounding storing customer voiceprints in a database have meant that some organisations are worried about the potential backlash of trying to implement these technologies. The media have often fuelled fears around data privacy breaches, meaning customers have historically been keen to hold tight to their information. But this tension is easing and businesses are beginning to see a change, where customer data is more readily available and privacy is becoming less of a pressure point.
April 2016
FEATURE The arguments against a customer having their voiceprints stored are diminishing, along with the obsession about privacy. Industries like banking are beginning to get over the fear of using their customers’ data, in turn allowing them to provide a more seamless authentication process through biometric technology.
Facilitating the change Governments also have a key role to play in embedding voice biometrics in day-to-day use, and encouraging consumers to accept the technologies. In the UK, a recent parliamentary report6 criticised the government for failing to have a strategy around biometrics. Governments worldwide need to work with external parties and experts in this area to inform a voice biometrics strategy. For example, the Australian tax office has recently announced plans to roll out biometric authentication applications to its mobile app7, in addition to the 1.5m voiceprints it already uses on a daily basis. It would be encouraging to see public services in other countries following suit and making wider use of voice biometrics to achieve cost efficiencies and drive out fraud, as well as allaying consumer concerns through normalising biometrics. The public sector has a unique role to play in encouraging mainstream adoption of this compelling technology.
Good cop, bad cop Of course, as voice authentication gets smarter, so do the criminals trying to break in. Criminals understand that it’s quicker and more efficient to buy card details from the dark web, rather than hacking through endless layers of security. The downside to buying customer details through this untraceable source is that the criminal has all the details, but no way of knowing how much money they have available to them in the account. This is where things start to get interesting. Fraudsters have on many occasions been known to pose as the customer and call into contact centres and ask how much money is in the account. In the past, contact cen-
tres have been giving the information away freely, allowing the fraudster to spend the correct amount of money. Now, using voice biometric technology, the contact centre can recognise known fraudster voices on even innocuous-sounding calls for balance inquiry. Upon recognition, the contact centre can alert the customer and fraud departments to shut down accounts before any money has been stolen, stepping in as a front-line fraud detection service before any fraud has taken place.
“Using voice biometric technology, the contact centre can recognise known fraudster voices on even innocuous-sounding calls for balance inquiry, alert the customer and shut down accounts before any money has been stolen” To beat these threats, businesses need to step up and embrace the new technologies that are available to them. It is time that organisations across the private and public sectors recognised that typed passwords – and the dizzying array of security questions that act as fail-safes – are reaching the end of their useful lives. Voice biometrics have reached a level of maturity and sophistication that the technology can provide a fast, painless and highly effective form of authentication – one that will help engage customers, as well as protecting them. In summary, voice biometrics is a weapon that organisations can use to battle and eliminate fraud and maintain credibility. Alongside this, businesses can be champions of customer service, ensuring they get their customers' queries answered quickly. It’s a win-win situation for everyone except the fraudsters!
About the author Amit Desai is the senior director of solutions marketing at Verint Systems. His professional career has included executive roles in products and marketing at voice biometrics, speech
recognition and analytics software companies in Silicon Valley.
References 1. Nick Clements. ‘How To Protect Your Credit Score When You Are An Identity Theft Victim’. Forbes. 24 August 2015 Accessed March 2016. http://www.forbes. com/sites/nickclements/2015/08/24/ how-to-protect-your-credit-scorewhen-you-are-an-identity-theftvictim/#7f17e5c43f18. 2. ‘New fraud figures show fraudsters directly targeting bank customers’. Financial Fraud Action. 2 October 2015. Accessed March 2016. http://www.financialfraudaction. org.uk/news-article.asp?genre=media&Arti cle=3019&page=1. 3. ‘Customer-Centricity: The Rules of Engagement’. Verint. Accessed March 2016. http://info.verint.com/ LP=2421?utm_source=verint&utm_ medium=press-release&utm_ term=customerengagementresearch&utm_content=uk-version&utm_ campaign=EIS_GMD_15_CEO_WP_ VRNT_CustCentEngage_07Oct. 4. Oliver Griffin. ‘Fraud cost UK £732m last year’. Economia. 19 January 2016. Accessed March 2016. http://economia. icaew.com/news/january-2016/fraudcost-uk-732m-in-2015-kpmg-says. 5. ‘Cyber attack cost TalkTalk up to £60m and 101k customers’. The Financial Times. 2 February 2016. Accessed March 2016. http://www.ft.com/fastft/2016/02/02/cyber-attack-cost-talktalkup-to-60m/. 6. ‘Current and future uses of biometric data and technologies: Sixth report of session 2014-15’. House of Commons Science and Technology Committee. March 2015. http://www.publications. parliament.uk/pa/cm201415/cmselect/ cmsctech/734/734.pdf. 7. David Swan. ‘ATO signs deal to roll out mobile voice biometrics’. The Australian. 22 January 2016. Accessed March 2016. http://www.theaustralian.com.au/business/technology/ato-signs-deal-to-rollout-mobile-voice-biometrics/news-story/ c348e56b8cf722746b1fe7a2bbc04117.
11 April 2016
Biometric Technology Today