Our client and the project
Groupe Mutuel is an insurance company with about 1.3 million private customers in Switzerland and whose focus lies on health insurance. Their client base growth resulted in increasing customer requests, which Groupe Mutuel aimed to process in part with a chatbot. This chatbot, however, wasn’t sufficiently harmonised with the actual user needs. We organised user research in two stages to assess the main pain points and suggest improvements.
Structuring Groupe Mutuel’s knowledge base and investigating chatbot limits
We supported Groupe Mutuel with fully remote research which aimed at closing the gap between user needs and the abilities of their product, allowing the chatbot to communicate clearly as a standalone solution. The research included two stages:
In the first stage, we structured Groupe Mutuel’s existing support resources that were available for the customers to read, such as FAQs. Through card-sorting interviews, we outlined which topics and issues the customers would group together. Then, we investigated how customer support could take on the more complex requests that clients were unable to solve with the help of said online resources. Each client picked a scenario including a support request, which they had to solve with the help of the Groupe Mutuel chatbot. The scenarios could be anything from “I have a question regarding a bill” to “which form do I need to ask for accident compensation”. We observed that many of the customers were unable to solve their problem with the help of the bot. Additionally, they did not know how to proceed to find their answers once the Groupe Mutuel chatbot had reached its limits.
For the second part, participants were put into small groups, so-called focus groups, to outline and discuss their general attitudes towards the chatbot, and the positive and negative experiences they had had up until then. The groups were made up of the people involved in the chatbot flow at Groupe Mutuel. These people not only included customers who had a question, but also the customer service employees who then took over from the chatbot. For most of the clients, their experience was marked by frustration, mostly because the interaction with the bot was not conclusive, and it did not offer any concrete solution to get in touch with customer service afterwards. For customer service on the other hand, it was their main goal to help their clients and keep them happy. A need the bot clearly did not meet.
Chatbots need to be understood, too
From our research, we were able to put together a report that outlined clear measures for Groupe Mutuel on how to optimise their chatbot so that it efficiently support customer requests. First and foremost, the exchange between the bot and the customer needs to develop smoothly. Particularly for the cases where the bot reaches its limits and is unable to answer a question, it needs to communicate this clearly and offer alternatives for the customer to proceed. The solutions to the customer’s request could be in form of further resources they can consult for themselves or the handing over to Groupe Mutuel’s customer service. Another optimization suggestion is to minimize the friction to get an appointment: Groupe Mutuel should aim to minimise the waiting time or give the client the option to pick an appointment to cut the waiting completely.
The conclusion of our research led to easy to implement and clear solutions for customer happiness and customer support relief, allowing a future chatbot to fulfil its intended purpose.