Get Ready for Chatbots! - Part 4
Harnessing the Full Potential of Chatbots
In the previous parts of our article series we have shown the many benefits of modern AI-based chatbots. However, we often observe that insurers experimenting with chatbots only use a small fraction of the capabilities that modern chatbots can offer. In the following we will show what design guidelines insurers should consider such that they can tap the full potential of chatbots.
Structured and Unstructured Input
It is important for chatbots to be able to process unstructured user input in the form of natural language, as this is the most intuitive way in which we communicate with each other. This allows customers to interact with insurers via chatbots just like they would when chatting with their friends.
Many insurers however design their chatbots in such a way that they still only can process structured input using input masks and menus – an approach that does not actually make use of artificial intelligence. Hence, they miss out on the improved user experience and self-learning abilities that modern AI-based chatbots can offer.
Is structured user input completely useless then? No, because there are also use cases where for example the users’ answer options should be limited or where it is more convenient for users to enter information via drop-down menus, sliders etc. than via natural language.
Optimal Customer Experience
We therefore suggest insurers to use a well-considered combination of structured and unstructured inputs to guide customers through their business processes as user-friendly as possible.
Some additional methods that insurance companies can use to achieve optimal customer experience include: using autocomplete features and predefined default values to make the user input more convenient. Allowing the user to send and receive pictures, videos and voice messages where writing or reading a text message takes too much effort. And finally, integrating external services such as map or payment services via APIs to allow users to quickly enter and send information.
Flexible Conversation Flow
Imagine that in order to go through a certain business process of your insurer, you would have to answer dozens of questions one by one from one of your insurer’s forms in a fixed order via a messaging app. You don’t think that this sounds very appealing? Neither do we – but unfortunately, this is the status quo of many chatbots that insurers are experimenting with today.
Hence, chatbots should allow for a non-linear conversation flow where the user do not have to follow a pre-determined dialogue path. The customer should be able to describe his personal situation to the chatbot in a flexible way and ask questions in between (similarly to how he would to a human insurance agent) while it fills out the form for him. The chatbot’s main task is to guide the customer through the business process as efficient as possible and to assist him where necessary.
Flexible conversation flows can easily be implemented using AI-based chatbots since they use an intent-entity-based system instead of rulesets. Still, profound knowledge of the business processes is required to successfully create the links between the dialogs, intents, and entities and to avoid pitfalls such as getting stuck inside a loop.
Interacting with insurers is most probably not among your top things to do in your spare time. This is due to, among other things, the many questions you have to answer and the tedious steps you need to take to go through their business processes. To improve the customer experience, insurers should keep the processes lean and only ask for information that is absolutely necessary. Additionally, the chatbot dialogues should be written as comprehensible as possible such that customers not familiar with insurances can also easily understand them. Finally, the processes should be automated as much as possible to reduce processing times.
As you can see, modern chatbots offer vast capabilities that should be fully harnessed by insurers. Yet despite their many benefits, chatbots can still fail at gaining customer acceptance if insurers use the wrong strategy to incorporate them into their customer interaction. We will address this challenge and provide a solution in the next part of our article series.
You might also be interested in the following articles:
Are you curious about trying out our prototypes? Would you like to talk more about chatbots? You can reach us directly, using the contact details below.