How to detect emotions in a chatbot conversation?

Detecting emotions in a chatbot conversation

A sentiment & emotion analysis tool can detect the polarity of an exchange as well as the possible emotions present. It is a key tool for giving value to the information coming from customer interactions, especially when these exchanges take place with a chatbot.

What does your customer really think?

The digitization of customer relations implies replacing certain interactions between the customer and an advisor or a salesperson by exchanges with conversational agents or chatbots. This tool, when well-adapted to the company's activity and to the customer's needs, can accelerate and facilitate most of the processes. For the company that implements it, the chatbot saves time, increasing the efficiency of the teams of advisors who can generally concentrate on tasks with higher added value. It allows advisors, whether in pre-sales or in after-sales service, to limit their interactions with the customer to priority or complex subjects. However, this use of chatbots carries the risk of missing key information about the customer. Worse still, automation runs the risk of missing out on a customer's dissatisfaction with, for example, their telephone subscription, their annoyance due to repeated power cuts, their exasperation at not being able to find the information they want on their online banking application, etc.

Some human-machine interactions cannot be automated because they require empathy from a customer service department, an attentive ear and, above all, speed in processing requests. Unfortunately, sales and after-sales teams lack the resources and time to make use of this information in a timely manner. And very often this frustration with the chatbot can increase the risk of attrition (churn rate).

Obstacles to a Conversational Approach

• Fewer and fewer interactions with the customer on certain topics
• Loss of potentially key information for customer retention
• Little time for teams to analyze chatbot conversations
• Difficulty in quickly identifying the breaking point with the customer and act accordingly

NLP to enhance the Chatbot Interface

Gathering important information from all customer interactions and prioritizing actions to be taken is a challenge integral to the development of chatbots. To meet this challenge, Lettria has developed a sentiment analysis API that integrates with the conversational agent in order to improve its comprehension capabilities. To do this, our solution goes through two steps:

• An algorithm identifies the adverbs that allow the speaker to give a particular meaning to their words, and then to measure and to qualify them. After locating the opinion expressed in the text, the algorithm assimilates it into a source.
• Deep Learning then can link this analysis to a context and a framework that can give meaning to the feelings and emotions expressed.

By doing so, the chatbot is already better equipped to understand the customer's needs and intentions and to provide an initial response that is adequate in terms of content and form.

Sentiment analysis will then automatically analyze the interactions with the chatbot and identify the customers who need special attention. By polarizing the opinions of chatbot users, our API provides key information to sales and customer relationship teams to prioritize and better target their actions.

Our Solution's main advantages

• Ultimate goal is to reduce churn and improve the customer experience.
• Improve customer experience by increasing the chatbot's understanding capabilities.
• Improve prioritization of actions with the customer for pre-sales, customer relations and after-sales teams.

Discover our partners' current use cases

• Reduce churn in retail banking and insurance: Increasing the comprehension capabilities of a chatbot is crucial, especially with the rapid development of online banking where interactions with a bank advisor tend to decrease or even disappear. Capitalizing on interactions with a chatbot allows for the identification of the breaking point with the customer and the sources of dissatisfaction, and it allows for the mobilization of the right resources to build customer loyalty.
• Improve communication and user experience in the Transportation (train and aviation): Delays and incidents in transportation are a source of stress and tension for users, which greatly complicates the management of customer relations. Showing empathy is the key to untangling tense exchanges with chatbot users. Lettria's API becomes essential in order to enable the chatbot to efficiently process requests for information.
• In sectors such as telecommunications and energy supply, where interactions with customers are essentially limited to subscription or resiliation, it is crucial to capitalize on the rare interactions. However, more and more, these events begin with a chatbot conversation.

‍

‍