CASE STUDY: NSPCC
The NSPCC has used voice recognition for almost as long as it has been using direct-response TV ads. During those ten years it has used the technology to handle spikes in response so that they do not miss any potential donor calls. Fundraising manager, Sara Mansfield, describes how it works.
"Callers are presented with a range of options, the system recognises the response and the caller is then able to input bank details, leave a number so we can call them back, or continue to hold for an agent.
"The benefits are obvious in that we are able to answer a greater percentage of the calls presented to us immediately after our advert is aired. We aim to answer 90 per cent of calls using our expert bureau team and 10 per cent via the system at times of peak-call traffic. On average this is achievable." NSPCC used an agency to install the the system, and Mansfield advises anyone else to do the same in order to get guidance on how to balance the calls between agents and the system, to ensure that the agents are prioritised. She does not see voice recognition as a cheap and easy solution to DRTV call handling. "Although the general public is becoming more used to using these systems and so is more accepting of them, the response rates are still lower than if the call is answered live." She concludes that although the system has a place, it is no substitute for a good agent. "We've looked at the costs and responses via both call handling methods and discovered that while a call handled by voice recognition is on the face of it cheaper, the greater response and accuracy rate comes via agents, making that method more attractive."
Phone systems already enable charities to take every call at peak times, but new technology can also engage callers in conversation too, as Alex Blyth discovers
It used to be possible to recognise financial traders by the fact that they would spend their working day shouting down a telephone, ordering their brokers to buy or sell some commodity, bond, or stock. While some things never change, others undergo significant but barely noticed revolutions.
Today, traders at Citibank are still shouting their orders down a phone, but the broker has been replaced by a voice recognition system, a machine that understands the words and acts upon them. As this technology becomes ever more capable and more accepted by the general public, so those working in the not-for-profit sector will need to understand its potential benefits and dangers.
What is voice recognition technology?
There is considerable confusion over exactly what is meant by voice recognition technology. For example, one company, EasyLink, sells what it describes as a "voice-recognition solution". This is essentially an automated, pre-recorded voice message that is sent to a landline or mobile telephone.
The recipient can respond to the message using an automated response system. This can be a powerful marketing tool and one that would seem to be particularly well suited to the charity sector, given the receptiveness of donors and consumers at large to communication from charities, particularly at a time of emergency. However, while potentially useful, it cannot truly be described as voice recognition, as the system recognises keypad entries, not the user's voice.
While most consumers are familiar with systems that allow them to select options via a keypad entry, increasingly, voice commands are taking hold in the market. Enabling people to use their voice is faster, easier and more intuitive, and so has to be preferable in the main. Moving beyond this, some systems are able to cope with 'directed dialogue'. This means that they respond to certain key phrases or words. This is used fairly extensively in the banking sector where customers might, for example, be directed to say "current" if they would like to hear the balance on their current account.
However, voice recognition is really coming into its own with the introduction of 'natural language' speech recognition systems. These are open environments in which the machine is able to respond to anything, including open questions.
These systems, which are powered by elaborate algorithms, are complex and expensive, but they are becoming more popular in the commercial sector and will eventually become so on the not-for-profit side.
How is it being used in the sector?
At the moment, voice recognition is not widely used by charities at all.
Some enthusiastically point to the fairly widespread use of interactive voice response as evidence of an industry embracing the technology.
Interactive voice response is used to mitigate the difficulties associated with predicting the number of responses there may be to a fundraising television advert. Sara Mansfield at NSPCC explains: "While we aim to handle the vast majority of calls using trained, motivated bureau staff, we also do not want to lose any important fundraising calls should the spike of response be higher than we anticipated."
A system whereby donors can leave either direct debit details or a number for an agent to phone back means that far fewer calls are lost. While interactive voice response is undeniably useful, it is not really true voice recognition. To many people it is simply a system of "glorified answerphones".
Where charities are introducing voice recognition technology, it is being used primarily to automate switchboard services. The National Centre for Young People with Epilepsy recently introduced the technology to its switchboard. Fundraising officer Tina Challis describes how it works.
"We've got about 200 residents here, and our receptionist was deluged by donors, parents, social workers and others calling in. The new system was installed fairly cheaply and simply," she says. "It asks callers who they want to speak to, recognises the response and puts them through.
"It's made us much more efficient, meaning that we can focus on our care duties. I think it's probably impacted on fundraising by making us seem more professional and ensuring that every single call is answered," says Challis.
Don Edwards from IT consultancy, Fujitsu Consulting, believes that this sort of basic voice recognition technology could be adopted by charities for a wide range of tasks. "These simple systems could answer questions about the purpose of the charity, the location of shops and offices, or how to go about making a donation," he says.
And, as one vendor points out brutally: "With a one-off installation cost of £20,000, these systems are pretty good value when compared to a receptionist."
However, charities are yet to tap into the full potential of this new technology. Using voice recognition in a limited way is akin to using your laptop as a calculator.
The RNIB is looking to take it a step further, as digital policy development officer, Julie Howell, explains. "About 11 per cent of visually impaired people in the UK have PCs in their homes, compared to about 98 per cent with phones. Furthermore, not all blind people are fully competent in Braille, making sound and voice even more important to them. So, we're working on ways for them to interact with our newsletter and product catalogue via the telephone," she says. "Put crudely, the voice replaces the mouse and keyboard, and hearing replaces sight."
The RNIB knows that this model is technologically possible, and over the next two years will be exploring how to make it practically and financially viable.
The people problem
While the RNIB is more forward-looking than many others in the sector, its plans are still conservative. Voice recognition technology has much greater potential, and the low take-up to date is a reflection of the many issues it has faced in its short history. The most frequent criticism is that people want to speak to people.
Karl Holweger, chief executive at telephone fundraising agency, Pell & Bales, believes that it is a "technological solution of last resort that is often more useful to the organisation than it is to the customer".
"When most people make a call they expect to speak to another person and they don't expect to be shunted to a computerised system," he says.
"It can be useful to cope with unexpectedly high responses from TV fundraising ads, or to provide a '24/7' service, but beyond that I would be really wary."
The charity sector has a great deal to lose in terms of reputation and goodwill from inappropriately used technology. This has led one supplier, Intervoice, to set up a team dedicated to improving the quality of voice recognition systems.
Simon Edwards, marketing director identifies three key elements which are essential to get right. "People will never be delighted to speak to a machine, but they will accept it if you ensure it is appropriate, working and up-to-date," he says. "So, a charity system ought not to be too ostentatious.
The system should be quick and should have no dead ends, and, like websites, it will only be used if it is regularly maintained and updated."
Many commercial organisations shy away from voice recognition technology once they discover that maintenance and support expertise is almost entirely on the vendor side. The lack of available in-house expertise is also a concern for charities. However, as the market matures and more vendors are lured client-side, this will become less of a problem.
In the same way, the final problem, cost, is less of a barrier than it once was. Steve Desay, principal consultant at contact centre consultancy, Catalyst IT Partners, estimates that a basic system might cost as little as £5,000 to buy and set up. And while a more complex system will cost up to £150,000, most in the industry agree that a well run system will produce a return within six months.
Furthermore, charities, particularly smaller ones, may be attracted by hosted services. These are paid for on a usage basis and so are more financially attractive. Another piece of good news is that Microsoft is entering the market and so prices for the more basic systems and applications will almost certainly be falling in the not-too-distant future.
There are undeniably problems and the really clever technology is still beyond the reach of most charities. However, the future is always closer than we imagine and it would be a foolish charity that chose to ignore the benefits, summed up here by Desay. "A system that can resolve transactions of any type - be they information retrieval, payment handling or call routing. It can do all of that by interacting with callers via the spoken word. It can do it 24 hours a day, 7 days a week. It never goes on holiday, is never ill, and, if you set it up right and look after it, it always does what you want it to do."
Voice recognition technology describes a system which interacts with human speech. Simple versions include those where the caller chooses a word from a limited range of options and the system then acts on that choice. On more complex systems, callers can interact through speech with a programme that learns as it goes along. Key vendors include:
Intervoice has led the voice automation market for the past 20 years and is now aiming to deliver high-quality voice recognition systems.