If you’re one of the 1.2 billion or so Facebook users who have already downloaded the company’s Messenger app, then you might already be immersed in the brave new world of bot-enabled services.
The social media behemoth began phasing out its existing mobile chat facilities last year to make way for a new platform that would also support third party chatbots. For Facebook, an expected proliferation of bots represented an opportunity to monetise its chat channel by enabling commercial partners to engage with the public via automated text messaging, underpinned by artificial intelligence. Meanwhile, the partners themselves—those deploying the new virtual assistants—had an opportunity to develop next generation customer services.
Since 2016, Facebook’s march towards the bot-enabled future has been relatively brisk. In April of this year, the company announced the opening of a bot-store—to make it easier for users to find useful services—plus extensions to the platform, enabling bots to be incorporated into chat sessions between friends or within groups.
Facebook’s initiative reflects a bigger commercial trend. From the perspective of a company seeking to keep a check on costs while upping the ante on customer service, the idea of an automated chatbot that can answer questions, provide instruction or guide a customer through an e-commerce process is hugely appealing, not least in financial terms. Engaging with customers via a bot is, after all, a lot cheaper than hiring contact centre staff, at least in the longer term. But the real opportunity lies in developing innovative new ways of serving customers rather than simply automating existing call centre functions.
What are the use cases for bots?
Bots can of course be used as an alternative to calls or live chat. For instance, travel company Cruise 1st currently uses a chatbot to handle low-touch customer inquiries. “Our bot can handle simpler queries, allowing our human agents to deal with the more sophisticated and challenging ones,” says Simon Hoe, Global Head of Digital and E-Commerce.
But cost-effectiveness is only one small part of the use case story. A well-designed chatbot can be a catalyst for the provision of added-value services.
A bot launched by credit score agency, ClearScore is a case in point. Essentially the bot is designed to offer customers advice on how to build, repair or improve their credit scores. “By educating them via simple tips and advice in a format they could engage with, we hoped to give consumers the best chance of securing the best possible financial deal,” says Anna Kilmurray, the company’s chief marketing officer.
Bots can also be deployed to humanise the interaction between man or woman and machine when tasks are being carried out online.
Witness Peg, a bot developed by online accountancy company Sage. Available for both Facebook Messenger and Skype, the tool provides a chat-based interface between the user and Sage’s software. Given that Peg uses natural language, Sage subscribers can simply ask (via text) questions such as “how much have I earned this month” or “who owes me money” or carry out tasks such as filing expenses (along with attached pictures of receipts).
As Sage’s VP of Bots and Artificial Intelligence, Kriti Sharma explains, Peg’s interface provides a means for customers to take advantage of artificial intelligence systems developed by the company. “Sage has been developing machine learning and AI specifically designed to reduce financial admin for business builders,” she says. “When AI is used for set, repetitive tasks—for example, accounting—it has the potential to significantly reduce admin.” In other words, Sage has developed Peg to make the task of accounting simpler and easier.
What data bonus can bots offer?
Meanwhile, with analytics systems running quietly in the background, a bot can also provide the kind of customer data that is difficult to capture via more conventional means. A contact centre CRM system captures the main purpose and outcome of calls—from problem through to agreed solution—but not the nuances of the conversation. Calls can be recorded in their entirety, but audio tends to be reviewed only when a problem arises. It is not a good medium for analysing and comparing engagements to, say, identify recurring problems. Text based conversations, on the other hand, can be searched and analysed relatively easily.
There is also an opportunity to make customer service much more data driven. ClearScore’s bot enables advice to be tailored to individual credit profiles. “The bot uses the ClearScore database to instantly access key information from ClearScore users’ credit reports,” says Kilmurray. “This insight allows us to give tailored responses and advice designed to help any individual improve their credit score.”
Ajay Vij, Senior Vice President, Industry Head, Financial Services, at Infosys, says data collection and analysis is set to be a key driver for the roll out of bots. He cites the banking and financial services sector as an area where virtual assistants could be a catalyst for a service revolution.
“In 2014, a study by CGI found that 52% of customers wanted banks to show them how to save money and 55% wanted access to wealth-building advice by leveraging their financial data,” he says. “That’s becoming a reality. Banking chatbots use machine learning to understand customer behaviour, track spending patterns and tailor recommendations on how to manage finances.”
What obstacles blight the path of bot adoption?
But will the revolution really take place. Chatbot technology can be presented to the consumer in a number of ways, notably via a company’s own website, as part of the functionality of a mobile app or through a platform such as Messenger.
In the case of the Messenger model in particular, it is very much an opt-in medium. Customers actively choose whether or not to fire up a bot. If they are to gain traction, the experience has to be good.
Clearscore’s Anne Kilmurry acknowledges creating a system that customers want to use was one of the biggest challenges that had to be addressed. “We spent a huge amount of time getting the tone of voice for the bot right. ClearScore prides itself in talking to our users in a friendly, natural way. Our bot replicates that tone of voice and even extends it by using things like light-hearted GIFs as part of the conversation,” she says.
Kriti Sharma sees Sage’s users warming to their new bot friend. “We have started to notice that users actually form a relationship with the assistant,” she says. As Sharma sees it, customers see Peg as something more than—and different from—a conventional app. They interact using natural language and feel comfortable asking questions. “We actually see that some people prefer to ask the assistant questions, rather than their colleagues, because its less embarrassing to ask a dumb question to AI than it is another human,” she adds.
How about the thorny issue of ‘uncanny valley’?
There is, of course the danger that bot interactions stray into ‘uncanny valley’ territory, the robotics term for machines that are almost, but not quite, human—and therefore just a bit creepy. Morgan Lloyd of Welsh Water doesn’t think that’s a problem. His company used a bot (via Facebook) to carry out a legally required water industry consultation. The bot provided a more interactive means to talk to customers than, say, a questionnaire and about 3,000 customers took part. “I think most people do know that they are speaking to a bot and they know how to speak to a bot,” he says. “But it’s important that when the conversation goes beyond the bot’s programming that it will say something like, ‘I don’t understand’.
This raises the question of what customers expect from bots. Are they mere utilities that allow us to access simple advice? Or do we want an experience that more or less replicates a conversation with a real human being? More fundamentally, given that most of us don’t like talking to robot voices when we pay utility bills by phone, will consumers be any more comfortably taking financial advice from the new generation of smart customer service assistants?
Ian Bradbury, CTO of financial services at Fujitsu, says wide acceptance is still some way off. “Some people might find it very uncomfortable to talk to even advanced bots, though they are frequently using other digital channels,” he says. As such, he argues that bots should be offered as an option rather than a replacement for existing touch points.
But as Sharma points out, the idea of the virtual assistant is already accepted by many. “The boom in consumer-facing AI, such as Amazon’s home assistant Alexa, has helped the rate of adoption,” she says. And as systems become more sophisticated, bots underpinned by machine learning offer a means for companies to learn more about individual customers and to provide personalised interactions that can be delivered cheaply through automation.
This story, "Welcome to the brave new world of bot-enabled services" was originally published by IDG Connect.