The introduction of Amazon Echo and Google Home is helping the Chatbot crossover into the mainstream, heralding a new era for the technology. However, while they have grand ambitions of offering 360-degree personal assistant support to consumers, the current limits of the technology mean they still fall short of being able to complete more complex tasks. Sure, Alexa and Siri can tell you the weather, play your favourite Spotify playlist, search the internet on your behalf, and manage your diary – but ask them to book you a ticket on the next train home, order an Uber, or tell you your bank balance and they’re stumped. But, why?
First off, most chatbots are designed to be generalists, offering support on a wide range of tasks faster and cheaper than a human executive assistant. The problem is that these ‘virtual assistants’ don’t have access to the information they need for the provision of more complex support. Open data is fine, but when it comes to proprietary data like medical or banking information, they simply don’t have the permissions. Moves towards open banking are on the horizon, opening up the opportunities for bots to offer these services.
Then add to this that Alexa, Siri, Cortana etc. are in essence, conversational tools. Yet the current gaps in AI and machine learning mean that chatbots are simply not advanced enough to offer frictionless conversation. Designers have come up with ways of dealing with this, by flagging which words to use for the best outcomes, or providing multiple choice answers to a customer’s question, but often this limitation leads to user frustration, and encourages them to ‘opt out’. That said, despite the weaknesses of this disruptive technology in its current form, it is still possible to garner commercial value from chatbots.
The data processing power of Artificial Intelligence and machine learning is formidable – but applying this to too many things at once makes it difficult to be effective. So, those brands that develop chatbots for a specific purpose, and effectively integrate natural language input and dialogue trees, can create virtual experts that lead the conversation and provide a human-esque touch. Customers don’t always know what they want, or the right questions to ask, but an expert bot can use its knowledge to provide tailored and personal recommendations. This improves user experience, and ultimately unlocks new revenue and cost-saving opportunities for businesses.
Healthcare is a sector that has historically struggles with customer engagement. Using expert chatbots could allow providers to offer individuals their own personal healthcare assistant in their pocket. The assistant would be capable of giving advice and expert help as well as of automatically escalating the response and calling for human intervention, when needed. Look at HealthTap, for example, a startup that uses Facebook Messenger to connect patients with 100,000 doctors. The bot stores responses previously given to patients by their doctor, and can deliver the same advice to other patients with similar questions. If it can’t find a suitable answer, it connects the patient to a doctor directly. This makes healthcare delivery more cost-effective by reducing time pressures on doctors, whilst offering convenience and reassurance to patients, simultaneously.
Professional services is another industry where expert bots could be put to good use. Last year saw the birth of ‘the world’s first robot lawyer,’ in the form of a free service called DoNotPay. The chatbot has won 64% of the cases it has taken on, overturning almost £4m in parking tickets in London, New York and Seattle. By asking a set of simple questions – like were there clearly visible parking signs – the bot first works out whether the appeal is possible, and then guides users through the appeals process. On a wider scale, DoNotPay demonstrates how law firms (and potentially others including financial advisers, banks, accountants and insurers) could utilise chatbots to open up new revenue streams, by efficiently delivering services to a new segment of clients at affordable rates.
One UK startup, Cleo, has already clocked onto this, and is using an expert bot to offer personalised financial advice to consumers. With ‘read only’ access to the customer’s financial data, Cleo can collate a high-level view of their ingoings and outgoings. It is then able to analyse their financial situation, provide insights and answer questions via SMS and Facebook Messenger. The chatbot can also take this expertise one step further; with permission from the customer, Cleo can automatically shift money left over at the end of the month into a savings account. The potential for this is huge. Using similar technology, retail banks could cost effectively offer one-on-one financial advice to millions of customers, for example. Not only does this boost customer satisfaction and reduce churn, it could also present cross selling and upselling opportunities.
AI and machine learning will inevitably become more sophisticated in the years ahead. However, for those looking to use these technologies to support customer experience now and in the near future, it is important not to shoehorn the technology into something it is not currently able to deliver without causing friction in the customer journey. Creating specific, expertise-focussed chatbots utilises the strengths of the technology, enabling them to lead the conversation, rather than putting the onus on customers to use the ‘right’ words or phrases. By working with the natural limitations of chatbot technology, fully supported by proprietary customer data – and additional data collected over time – businesses will be able to adapt to provide virtual, on-hand, task experts. This will, in turn, enhance the customer journey and unlock new revenue opportunities.
Sourced from Oliver Shreeve. senior design consultant, EY-Seren