Introduction
Faced with a high volume of advisor-client conversation records across the region, CIC Sud-Ouest identified a key business challenge: many clients were being mailed or contacted with offers they had just discussed with their advisor, generating inefficiencies and risking client fatigue.
Influence Metrics was commissioned to analyse hundreds of thousands of advisor notes and text records using advanced NLP (Natural Language Processing), LLM (Large Language Models) and text-mining techniques. Our mission: extract meaningful themes from text, map them to individual client IDs and deploy intelligent workflow triggers, so that offers are relevant, timely and aligned with what was actually discussed.
Challenge
- CIC Sud-Ouest manages thousands of advisors and manages large volumes of unstructured text data (advisor visit notes, CRM entries, client IDs).
- The bank needed to ensure marketing-automation campaigns did not redundantly target clients whose needs had already been touched in conversation.
- They sought to segment clients by real, contextualised needs (e.g., auto insurance, credit, card issuance, account opening) rather than by high-level demographics or old behavioural scores.
- They required full data-governance, confidentiality and compliance with banking standards.
Our Approach
- Data ingestion & preparation
- Aggregated hundreds of thousands of textual records from advisor systems, mapped to client IDs.
- Cleaned, de-identified and structured the data to ensure confidentiality and GDPR compliance.
- Text mining & semantic analysis
- Applied NLP pipelines and LLM-based semantic models to detect and classify themes discussed in each interaction (e.g., “auto-insurance”, “credit request”, “international card”, “account opening”).
- For each client ID, created a profile of topics covered and topics pending.
- Workflow design & activation
- Designed a rule-based engine to trigger marketing workflows only when the client had not recently discussed a given topic with their advisor.
- Integrated with the bank’s CRM/MA system to ensure offers were personalised and timely.
- Dashboard & KPIs
- Delivered an interactive dashboard for CIC Sud-Ouest showing topic-coverage by client segment, number of eligible clients per topic, and conversion potential.
- Provided training and documentation for internal teams to use and monitor the solution.
Results
- A marked reduction in redundant offers sent to clients who had recently spoken to their advisor, enhancing client trust and marketing efficiency.
- A higher relevance of offers: marketing communications corresponded to actual needs identified in the advisor-client dialogue.
- Improved time-to-activation of follow-up offers, since the text-mining output directly fed into automation workflows.
- The bank strengthened its position as an innovator in the regional market by adopting AI and LLM-based client-insight modelling.
Conclusion
Through its partnership with Influence Metrics, CIC Sud-Ouest has turned what was once unstructured text noise into a strategic asset. By applying advanced NLP and text-mining at scale, the bank now offers more relevant, efficient and respectful communications aligned with each client’s true lifestage and conversation.
This case underlines how data-driven text-analysis can bring measurable business value in the financial-services sector.
