How Solimut Mutuelle de France Turned 500,000 Protected Members into an Intelligence Source for Service Quality Management

Solimut Mutuelle de France provides supplemental health and life insurance coverage to more than 500,000 people, with over 60 agencies across southeastern France. Aurélie, Head of the Voice of the Customer department within the Member Experience division — created in 2025 — oversees customer listening across all touchpoints and journeys: post-call, post-agency visit, post-reimbursement, subscription, cancellation, and external reviews via Trustpilot and Google.

Before Feedier, Solimut relied on a traditional semantic analysis solution. The result: keyword detection, not sentiment analysis. Verbatims categorized without context. No way to measure how a topic evolved over time after a corrective action was taken.

Aurélie was looking for a platform capable of structuring member knowledge around categories specific to Solimut, making sense of verbatims, and — above all — concretely measuring the impact of actions on satisfaction.

What convinced Solimut was Feedier's ability to go far beyond keyword recognition. LLM-powered semantic analysis captures meaning, not just surface-level patterns. Topics are fully configurable to match the business context. And time-based tracking turns every action plan into a measurable hypothesis.

Data Protection

In the mutual insurance and health insurance sector, data sensitivity is a key concern. Solimut Mutuelle de France protects the health data of more than 500,000 members, and compliance is non-negotiable. Feedier meets these requirements: the data collected remains isolated, hosted in compliance with European standards (GDPR), and is never used to train third-party models. For the Solimut teams, this guarantee was a non-negotiable prerequisite in choosing the platform. It enables the deployment of customer listening on a large scale without compromising member confidentiality.

Building the topic plan required genuine iteration — and that process is precisely what produced a reliable foundation. Four concrete challenges were encountered and resolved.

Over-tagging. Topics that were too broad captured too many verbatims, making analysis unusable. The fix: tighten the instructions so each topic is precise, even if it means temporarily accepting a higher rate of untagged verbatims.

Untagged verbatims. Instructions that were too restrictive left 30% of verbatims without a tag. Three successive iterations brought that down to 18%, with a target of 12%.

Cross-topic confusion. The "simple vs. complex process" case illustrated the importance of neutral naming: a topic labeled "simple process" only captured positive verbatims. The solution was to rename it to a neutral label — "reimbursement process" — and let the instructions define what's included and excluded.

Internal jargon vs. customer language. Members don't use industry terminology. The instructions bridged the gap by listing the natural phrasing that customers actually use.

Solimut can now track how its key topics evolve over time, cross-reference sentiment with NPS, CSAT, and CES scores, and produce automated reports that operational teams can act on. What used to be a manual review of scattered verbatims has become a continuous management system. The Member Experience division now has a structured foundation on which to anchor every journey improvement decision.

For Solimut, Feedier is not a reporting tool. It's the infrastructure that transforms thousands of member verbatims into structured, iterable, and actionable business knowledge. The quality of the topic plan is not a technical parameter — it's the foundation of the entire Voice of the Customer approach.