There’s a chatbot for everything these days. Chatbots that do your online shopping. Chatbots that search document libraries. Even chatbots that advice on your wardrobe. The advancement of artificial intelligence and burgeoning uses cases are the reasons why many enterprises are taking them seriously.
According to McKinsey and Company in a November 2018 survey, 54% of 2,128 respondents from all enterprises mentioned that they are piloting virtual assistants solutions within at least one of their business functions in their organisations.
As chatbots start becoming a norm, expectations on the speed and scale of service will rise. Delays will be frown upon. Failings will be catastrophic. Increasingly for time sensitive, service intensive industries like banks, hotels and telcos, how good chatbots are hinges on how well they can understand human interactions and intentions.
Accessing bots is not a natural thing for some not all. How can bots be made with diversity in mind would probably one of the biggest opportunities for enterprises and solution providers.
To seize this opportunity, chatbots need to shift from being language centric which focus more on a user’s command of language towards being communication centric which focus more on a user’s preference of input (eg. voice, images, video) in order to serve a large portion of end users especially in markets where cultural and demographics are dynamics at play.
Useability is as important as usefulness
Chatbots and virtual assistants which are communication centric would need to consider usability in its full entirety. A few considerations could be firstly, the ability to recognise common nuances like cultural slang and accents with accuracy would be particularly necessary for countries and industries where they exist.
Secondly, the ability to predict requests and personalise recommendations on the go, especially for the fast paced mobile horde.
Thirdly, the ability to integrate with social platforms that users are accustomed to for example Facebook, Whatsapp, Twitter or dominant Asian chat platforms like KaKao and WeChat would also be a key consideration.
Connected not to a single information base but a knowledge fabric
Haven’t we all encountered chatbots that are dull or imprecise?
The reality that we hate to admit is that chatbots are only as good as the brain it is connected to.
For customer service to be effective, chatbots need to prepare for complex human questions that require a myriad of information that are interconnected. In the Singapore context, a telco customer might say to a bot “My screen cracked. Can repair? Don’t know under warranty.”
This question requires the chatbot to read between the lines, understand the situation by drawing information from multiple databases (eg. member’s CRM and authentic parts log) and offer a very personalised and relevant recommendation one would expect from a well trained customer service officer who anticipated the next question.
With global business complexity creating more information silos and rising customer expectations, the need to build chatbots and virtual assistants to be more sensitive towards human interactions becomes imperative. Organisations will need to focus on a communication-centric mindset, understanding customers’ usage behavior and most critically connecting their bots to an organisation’s knowledge fabric in order to build inclusive and intuitive chatbots.