[Proxem Studio] Jean-Rémy Dudragne’s interview, Customer Experience & Innovation Manager at Engie

Discover how Engie was able to implement actions to improve Customer Satisfaction and reduce attrition, thanks to the Semantic Analysis of multi-source customer verbatim.

Can you introduce yourself and your main missions at Engie?

I work within the ENGIE entity in charge of the sale of electricity, natural gas and related services to private customers in France.

As Head of Customer Experience & Innovation, I am responsible for improving the experience delivered to our customers. To do this, I lead four teams to identify how Engie can deliver the best customer experience and am responsible for coordinating improvement projects across the board.

The team of Customer Satisfaction Managers operates the systems for measuring Customer Satisfaction and collecting the Voice of Customers, and then distributes them widely within the entity. The Marketing Research team conducts surveys to collect changes in our customers’ expectations and adapt the content of our offers accordingly. Customer Journey Project Managers intervene across the board to resolve customer irritants throughout their journey. Innovation Project Managers identify opportunities for innovation, in terms of customer relationship channels, new products or services that can improve the Customer Experience and then implement them.

Why did you implement a Semantic Analysis Solution of your Multi-Sources Customer Feedback?

We have a long history of hot satisfaction surveys. Following an interaction, we send a questionnaire with an evaluation, a comment, a free field note etc. In this way, we give the customer a voice as much as possible so that they can express themselves on the quality of interactions and services with Engie. We are therefore required to handle a very large volume of data and customer comments.

However, being able to analyze this data is essential to allow us to understand the reasons why a customer is satisfied or dissatisfied in order to act and re-engage the customer.

This is the main reason why we have implemented a multi-source customer verbatim analysis solution: service surveys, inbound channels, website, at the end of the customer journey.

Why did you choose Proxem for your projects?

We selected Proxem after a call for tenders mainly for two reasons: functionality and evaluation score. The platform offers multiple functions: search, filter, sorting of comments, representation in the form of a treemap, etc. They allow us to have a richer analysis compared to a traditional classification engine.

Through the various surveys allowing us to analyze the entire customer journey, we wanted to classify the verbatim according to the stages of the journey and the types of irritants.

We also wanted to have the same semantic classification for the irritant expressed in a comment, regardless of the channel. The solution is designed to be multi-source and with a common dictionary. That is, having a single classification plan for all sources analyzed.

In addition, Proxem was the only one to offer us a classification rating score. We wanted to maximize both the number of classified comments and have the most relevant classification result possible. The F score measure combines these two dimensions.

Could you describe the results of our solutions?

The analysis of multi-channel customer verbatim using Proxem Studio allowed us to:

  • Automate the extraction of the main themes emerging from customer verbatim;
  • Group verbatim dealing with the same subjects in order to identify risks and highlight weak signals;
  • Identify the levers of satisfaction and dissatisfaction distributed by customer typology.

We were thus able to democratize and make known what were the main irritants of the Customer Experience, on each of the reference customer journey. We were also able to be more relevant in the implementation of the actions that we proposed. Proxem Studio has enabled us to improve our overall Customer Satisfaction thanks to the adaptation of the Engie offer.

What are the perspectives for development?

We always aim to improve the quality of the semantic classification as much as possible to increase user adoption and have a more relevant result.

We plan to analyze speech-to-text data from telephone conversations, in addition to written Customer Feedback.

At the same time, we are going to analyze customer opinions on social networks through daily monitoring of the web.