How to Build a Private ChatGPT Using Open-Source Technology? Download our free white paper.

La Poste: Analyzing online reviews to improve customer service


Ready to go for the gold with Lettria?

Analyzing online reviews to improve customer service

La Poste Group is historically present throughout France and responsible for its postal services. Over the past few years, its services have diversified widely, from banking to cell phone operation, as well as the provision of digital services and insurance.

A key player thanks to its dense network of branches throughout the country, La Poste has more than 17,000 branches and post offices in France. On site, the advisors must respond to the various issues of the customers by ensuring a quality of reception and advice.

In order to analyze the feelings of its customers, La Poste mandated Lettria to analyze customer satisfaction in agencies through the comments they leave on the Internet.

How to use online comments to improve service quality?

For La Poste, the analysis of comments, ratings and online opinions is essential for maintaining a high level of quality in the reception and advice given to its customers. These comments, most often left after a visit to a branch, make it possible to estimate the feeling perceived by customers at each branch.

With more than 3,700 reviews and comments left per month, it is very complicated to qualitatively analyze customer feedback according to the branch they visited, as well as their expectations and level of satisfaction.

For La Poste, accurately analyzing these comments is essential. By better understanding its customers' expectations, the company can implement dedicated marketing actions, change its quality of service and measure the impact in real time.

Sentiment analysis as an indicator of customer satisfaction

Since June 2021, Lettria has been working with La Poste to analyze all comments left online.

Thanks to NLP analysis of customer verbatims, Lettria has been able to measure La Poste customer satisfaction based on the type of comments they leave.

For each comment, the content of the verbatims is analyzed in order to break them down morphologically (tokenization), to deduce coherent sets and establish relationships between them (pos-tagging and lemmatization). This first step identifies the main elements of the commentary.

Then a sentiment analysis is applied to extract the emotions and the tone of the customer according to each distinct element of the comment. Indeed, a customer may be very satisfied with his reception in the branch, but may have been disappointed with the price of the banking operations offered.

Finally, all the information is consolidated in a single platform (not provided by Lettria), to be used more easily by the business units. La Poste's network marketing department then has a simple and consolidated view of customer satisfaction and needs according to the type of request, its location and the affiliated product or service.


The next step in the partnership is the industrialization of the project, and its deployment to almost all agencies in France. New services are continuously integrated into the tool, such as the analysis of emotions in the comments.

Build your NLP pipeline for free
Get started ->