fbpx

Solving Chatbot Visibility Problem During the EDI Program

March 17, 2020
feelingstream
0

Feelingstream won the EDI (European Data Incubator) Program. During the last 8 months, our team had only one focus - to build an MVP for our Data Provider Telia. The MVP consisted of chatbot analytics and automatic reactions to various calls and chats based on our conversational analytics AI. Here are our reflections on the program.

Chatbot visibility and actionability

Every day we make decisions and use our full knowledge and experience to support those decisions. But how does one make decisions about improving customer satisfaction or prioritising customer service without true visibility and understanding of what the customers’ experience is like? You do it randomly - getting it wrong more likely than right. There is no way of telling which chatbot conversations are actually helpful, which situations are more troublesome for the bot and whether the responses given to the customers are even factually correct. All of these are small but important details and therefore usually invisible for the owner of the chatbot. 

In the EDI program, we accepted the challenge and built a conceptual solution for solving that problem. We constructed a chatbot flow discovery tool similar to process mining solutions that helps the manager of the chatbot track the specific details for every single path in the conversations. This gives the chatbot owner the power to know what is going on in the interactions and where the chatbot is failing to meet expectations, providing the same level of CS transparency in a company with 1m clients as you would expect to see in a company with 100 clients. Moreover, when critical signs are detected in the analysed chats, customer service agents are alerted who can react immediately. Actionability is crucial in the analysis platform to react fast and mitigate the bad experiences for the customers. 

Team reflection about the EDI

As a team, we do a lot of reflection among us to improve the development process. Here are our thoughts:


  • The new rigidity with set KPIs and a status table introduced by the program to the development processes made tracking the progress easy and transparent;

  • We were required to do more analysis for this MVP than we are used to and it impacted our efficiency positively. We will do our best to integrate this into our processes;

  • We were lucky to have our Data Provider so close to us because getting feedback from the customer as soon as possible is imperative to build what the customer needs, not what you assume that they need;

  • We got to build a prototype for a specific functionality for the first time and gained valuable experience on how to quickly roll out a mockup that can later be developed into a fully functioning product;

  • The approach we took with the data from the CS chats was truly revolutionary, giving invaluable insights into the conversations and impressing both the Data Provider and the EDI committee. We were lucky to have great cooperation with our Data Provider and were happy with the chance to improve their customer service yet again!


EDI wrote a story about us https://edincubator.eu/2020/03/12/the-story-of-our-startups-meet-feelingstream-one-of-the-8-top-edi-startups/, take a look!

EDI_winner_Feelingstream_Terje_Lauri_2020.jpg

EDI_winner_Feelingstream_Terje_Lauri_2020.jpg



The next phase in EDI - evolve - sales and marketing

The uniqueness of the Feelingstream solution is to find sales potential from existing conversations and improve the level of customer service. These can be calls or chatbot conversations, but the speed of acting with critical business signals is crucial to win the deal. 

There are now all doors open for us by EDI to go public.

Hey, like this? Why not share it with a buddy?

Related Posts

Leave A Comment