The ID callers can’t leave at home

Voice biometrics is here, it’s powerful, but it’s also dreadfully annoying, complains Philip Padfield

It’s amazing what technology can do to make the world a safer place – from identifying unique DNA, to locking out all but a single user from sensitive IT systems and accounts through the use of the latest biometrics. But isn’t this trying to crack a nut with a rather large jack-hammer in the typical call center scenario?

The argument is that, without sufficiently discretionary identification checking, contact centers might unwittingly disclose sensitive account information, such as a £250,000 bank balance. Not that interception of the Royal Mail or a quick rummage in the bins round the back of the local building society couldn’t achieve much the same thing…But is this really a winning use of intelligent speech recognition, or a symptom of a world gone a bit paranoid?

Some in the intelligent voice industry would have us believe that voice biometrics will be the catalyst for wholesale adoption of speech technologies. The logic goes something like this: Speaking is natural and effortless, and has the added advantage that your voice is something that is unique to you. Unlike other forms of biometric verification such as fingerprints or iris scans, it requires no special equipment to monitor and process it, and can be used over any phone, even over the Internet.

What’s more, unlike PINs or passwords, a caller’s voice cannot be stolen or lost, and it’s something they always have with them, so no danger of leaving it in the car or on the kitchen table…

A seminar we are running on the subject at Call Center later this month delves deeper into the issue and considering whether the time is right to be jumping up and down and whooping at this great next step in the fight against crime.

It will warn that, while the technology is good and ready and shows immense potential, organizations must be wise in the way they approach and deploy it. Solutions that require tedious ‘coaching’ sessions as customers ‘train’ the application to recognize their voices, may demotivate customers by presenting them with yet another hurdle they have to jump through to get decent service from a call center.

At the same time, as long as ‘false accepts’ lurk at the 1-2% mark on voice biometrics, which they do with many systems today, the technology is not safe to be used in isolation. Instead, it is best combined with standard identification and verification (ID&V) customer screening techniques.

These are all serious considerations for customer service managers seeking to drive the next wave of sophisticated and protection into their self-service applications. We’re not saying don’t do it; just make sure you implement a solution that works for and not against you.

Going backwards - Custom-build speech applications can create more problems than they solve, says Philip Padfield

With bad publicity over the general state of call center services, many organizations have turned to the most sophisticated technology they can get their hands on to improve the caller experience. Unfortunately, in the process, many have lost sight of their overriding aim (to raise customer satisfaction), and have ended up doing the exact opposite…

This includes trying to devise intelligent speech recognition applications that appear to try to delude the customer they are speaking to a live agent. This misses the point of the technology entirely, leading customers down a new road of discontent.

While custom-built speech applications allow companies to tailor the solutions to their exact requirements, they also provide unlimited scope to steer customers away from real, human interaction – the primary reason many people choose to try their luck at getting through to phone-based contact centers, rather than defer to Web FAQs and email!

To put too many hurdles between the caller and the agent is to tamper precariously with customer relations. This has become such a problem already that US consumer champion Paul English’s ‘Gethuman’ online database takes great pleasure in instructing customers how to bypass particular companies’ automation systems so they can get to speak to a live agent!

Speech recognition technology should be harnessed strategically and selectively, to facilitate better agent response, rather than replace it. Applied thoughtfully, it can transform the entire operation, boosting customer satisfaction and loyalty, and even driving up revenues as a proportion of call time is subtly refocused on upselling.

This means using the technology to filter out large volumes of routine, time-consuming calls, or substantial parts of them – distributing standard information, checking account balances, paying bills and so on – reducing waiting times for all. This frees up agents’ time so that they not only have a happier time at work, but can deal with more complex customer queries more efficiently.

Every company will have their own way of doing things, but if there is an 80% fit with a packaged speech recognition, this is the safest and most efficient way to go, knowing that the remaining 20% of the functionality you need that’s specific to your organization can be personalized quickly and efficiently. Reinventing the wheel for the rest makes no economic sense, and may lead the business down a blind alley - where original priorities get lost, projects derail and ROI is never delivered.

The end of software? ‘Pah!’ says Philip Padfield

The death knell for software has been sounded repeatedly over the last decade, yet, funnily enough, packaged software as we know and love it is a long way from kicking any buckets. Why is that, one wonders?

Indeed, for all the arguments in favor of hosted software services as a more sustainable alternative, there remains a wealth of support behind the ability to own, run, secure and manage the technology on an organization’s own premises.

Proponents of buying externally hosted applications on a no-commitment, pay-as-you-go basis, insist that this is the most cost-effective, flexible, low-burden way of accessing the latest functionality. And okay, this is often the case. Timothy Chou, the former head of Oracle On-Demand, in his controversial book,The End of Software, warns those of the old school that buying software outright incurs high costs whenever customization is needed, because this usually requires recourse to the greedy, conniving vendor. Conveniently, these expensive service fees must then be repeated each time there is a software update. A devious ploy, and no mistake!

In the contact center speech recognition market, this is understandably prohibitive, given the speed with which customer services are introduced and refreshed, and therefore system tweaks are needed.

The biggest mistake the software industry makes, though, is to force companies into a one-size-fits-all way of doing things. Customers should be able to choose how they procure and manage their packaged applications. What’s more, if they do choose to own and run these behind their own company firewalls, they should have the ability to change the various settings and scripts themselves, without the need for external help. That’s our line anyway. After all, we’re sufficiently financially robust without needing to line our pockets with ungainly proceeds…;-)

For the foreseeable future, the argument over whether organizations should buy or tap into their software functionality as a managed utility will be redundant. To talk about horses for courses would be to labor a hackneyed cliché, but that really is how organizations should approach the way they purchase software – basing their strategy on their own unique priorities.

While banks will have strict policies determining that anything touching customer data must remain firmly behind the internal firewall, other businesses may prefer the freedom associated with being able to buy into software functionality only as and when they need it.

In allowing all of the above, Fluency plays out the ideal scenario, in a way that other vendors can only look on with slightly yellowing tones... Not only do we support both on-premise and hosted software delivery models, we also make it easy for our customers to switch between the two, because the technology is the same whichever approach you take.

One of our US customers exploits this flexibility very well, using our technology on its premises for everyday customer services operations and then using additional hosted capabilities at peak times or for particular initiatives. Other companies exploit this flexibility to try out the technology before they commit to it.

Dealing with a vendor that supports both delivery models offers a further comfort factor for our customers, too, who are reassured by the fact that we are a well capitalized business that does not depend solely on the drip-feed revenue of hosted services provision…