Lessons learned around using virtual assistants and chatbots in customer centric care
Customer centric chatbots and virtual assistants
Despite initial hesitancy when they were first introduced, nowadays, on average, people really love their virtual assistants. So much so that, every year quite a few customers ask Amazon Alexa / Siri / Google Assistant / Cortana to marry them! While this might be funny, it indicates how the boundaries between man and machine are blurring. The virtual assistant is fast becoming an integral part of the family.
While many observers believe we are ahead of the curve in today’s dynamic landscape, the reality is that we are still in an artificial intelligence (AI) infancy stage. A lot of potential remains untapped with chatbots and virtual assistants – the best is yet to come!
Since many industry executives wish to pick up best practices, here is a breakdown of our key tips based on our successful experiences at Bajaj Allianz.
Do give voice and character to your chatbot
“Voice” here represents the tonality or the mannerism which you want to choose. While some firms wish to have a quirky and humorous tonality, other brands wish to keep it very formal and straightforward to prevent any risks of being misunderstood! The key here, based on our experience and research, is to choose the voice or character after a proper study of your target customer segment’s preferences.
Do not bury your customer with chat messages
One should consciously introduce wait time buffers between messages to give customers time to digest the conversation so far. A common mistake with chatbots is to load messages rapidly, which can be overwhelming for the customer. When messages are sent too quickly new messages will push the earlier messages (which the customers are still reading) out of the frame. A good example for wait times is PullString.
Do deliver hybrid customer support
This is one of the most important aspects in today’s world. Phygital – combination of digital and physical – is crucial. Seamless transition from machine to man enhances the quality of experience. Studies show that NPS on end-to-end resolution via a single action can be as high as 96, but if this resolution is split into two steps, the score can fall dramatically to 65. For instance, in cases where our chatbot, BOING, cannot answer a query, the customer is seamlessly transferred to a human agent who takes the case through to completion.
Do not allow the complexity of the back-end complicate the front-end
No matter how complex the backend (algorithms, data, permutations and combinations et al), the user interface of a virtual assistant or chatbot has to be convenient. In order to enable this, chat answers need to be short, crisp and simple. Users don’t want to scroll to read pages of indications in order to find the answers they need. Conversable is one of the classic examples on this front.
Do use quick replies on chat wisely
Using quick replies at the beginning of your chat conversations helps customers understand the types of queries the bot can cater to. These quick replies enable customers to get the servicing done quickly and then do a subsequent deep-dive into areas of interest.
Do train bots to handle fallbacks
Train the AI bot to identify the breakup points in the conversation and define an alternate solution to offer the users such as transferring the chats to the humans or call backs.
Do not fail to continuously improve
A key success factor for the success of any chatbot is continuous improvement which is achieved through continuous testing. Once the chatbot is moved to production, one needs to closely monitor interactions and collate feedback in order to understand how end users are interacting with a chatbot. In order to improve user experience, one can also use sentiment analysis to train chatbots on human capabilities.
Do value the power of aesthetics with user experiences
Users will enjoy using your chatbot even more if its design is aesthetically pleasing and representative of the brand’s image. You can add an avatar, a logo, colours, backgrounds, etc. A good example here is Hello Fresh’s Freddy.
As bots evolve and mimic humans even more, a future wherein bots and humans are equal partners at a worktable is not beyond one’s imagination. As Stephen Wolfram says, what will limit us is not the possible evolution of technology but the evolution of human purposes!