Accelerated Feedback Processing for Optimized Guest Experience

Challenge

As Pairi Daiza prepared for a significant increase in visitors, it sought to ensure that service quality and guest satisfaction remained consistently high. To support this objective, the park wanted to accelerate how visitor feedback—collected through satisfaction surveys—was analyzed and acted upon. The volume and variety of comments, often written in multiple languages and mixing praise with improvement suggestions, made it difficult to extract actionable items efficiently. A structured, scalable approach was needed to help teams identify key issues and deploy corrective measures more rapidly.

Nature of collaboration

Pairi Daiza partnered with Sagacify through the Tremplin IA program to integrate a generative AI assistant into Freshdesk, their existing customer support system. The collaboration focused on structuring satisfaction survey data, building a new tagging model to categorize feedback by topic and location, and classifying responses by type—such as complaint, remark, or praise. The solution was also designed to assist teams in preparing appropriate replies and streamline communication across internal functions, enabling faster interpretation and routing of feedback. Internal teams contributed directly to defining feedback categories and validating business rules, ensuring the system supported targeted follow-up and accelerated corrective actions within the park.

Solution

The implemented solution is a generative AI assistant integrated with Pairi Daiza’s Freshdesk environment, supporting the processing of more than 80,000 visitor messages annually. The system was trained to analyze multilingual feedback, identify dissatisfaction points, and associate them with specific areas within the park.

Key components of the solution include:

Automated extraction of complaint content from survey comments, with filtering of non-actionable or positive statements.
Standardized tagging across four dimensions: topic, subtopic, criteria, and location.
Classification of each message into defined types (e.g., complaint, remark, praise) based on sentiment and context.
Feedback routing to the relevant operational manager or department, based on the tags.
Assistance in drafting context-aware responses, aligned with internal guidelines and language preferences.

These capabilities allow teams to focus on high-value interactions while ensuring quicker follow-up and clearer visibility into service performance trends.