Siri, Alexa, Nina, Cortana, Facebook M, Google Assistant and others - chances are, you’ve used one of these. When Facebook announced in 2016 that it was opening its messenger platform to third party developers, the already inflated expectations around chat bots went stratospheric. The number of bots on Messenger went from zero to c.20k in five months*.
Chat bots allow communication with machines using natural language – voice, images or text. A number of global brands have piloted and launched their own bots or virtual assistants. KLM, for example, has logged more than 1 million messages since launching its Messenger chat bot in March 20161. Last year we helped one of our banking clients choose a multichannel vendor in order to realise their Digital Contact Centre strategy. One of their key requirements was the ability to implement a chat bot.
The popularity of chat bots is driven largely by the rise of smartphones and messaging. Smartphones have caused a shift from voice to text and images and messaging is being adopted at an unprecedented rate. The median usage of messaging apps is c.21 mins per day1. As familiarity and comfort increases, conversations are gradually moving from the social sphere to chatting with customer services about accounts and issues.
However, most chat bots today are limited to a very narrow set of scenarios even though they all claim to understand ‘natural language’. More often than not, they will force you to say or type things in a very specific manner before they understand you. The ability to have a free flowing conversation with a machine, like we all imagine, still seems to be the stuff of science fiction.
Having said this, they can and are playing a very useful role in an organisation’s suite of digital channels. Yes, they are no silver bullet (as they are often hyped up to be) but as they improve, they have the potential to make our interactions more natural and, therefore, easier. They are cheaper for the organisation and will soon have a positive impact on service, experience and ultimately, loyalty.
If you are starting to experiment, it is worth noting ironically, a successful bot is always backed up by human intervention. Start with the customer in mind and consider where the bot would fit into your ‘hierarchy of assistance’. Analyse your customer interactions to choose simple queries to begin with. In parallel, map key customer journeys to develop an in-depth understanding of each scenario in order to choose when and where to intervene or escalate. Starting with a pilot will allow you to improve the underlying knowledge base; further iterations can then involve sophisticated AI. No matter how sophisticated the bot, in order not to frustrate the customer, consider when human intervention is necessary and try to make it as seamless as possible.
The tide is gradually turning from apps to messaging. As advances in AI accelerate, chat bots will be able to better understand and respond to us. In the meantime, if you can, start experimenting as there is strong evidence to suggest that bots will continue to grow in prominence.
*Forrester: The state of Chatbots; Julie A. Ask, Michael Facemire, and Andrew Hogan, Oct 20 2016