A-262 Real-World Evaluation of Service Flow Redesign Using Proprietary AI Sentiment Analysis in a Clinical Diagnostics Support Channel

Abstract

The Contact Center is a strategic component of the customer journey in diagnostic medicine services, concentrating interactions related to tests, turnaround times, reports, and technical guidance. Until 2025, customer experience assessment relied exclusively on formal complaint records—an approach that is limited and poorly sensitive to users’ actual perceptions. With the implementation of the Ouvimos Você platform, the Customer Satisfaction Score (CSAT) survey via WhatsApp, and artificial intelligence–based sentiment analysis, it became possible to continuously and systematically measure customer experience. This study aimed to evaluate the impact of a set of operational, technological, and workforce management interventions on customer experience, using the Net Sentiment Score (NSS) as the primary outcome. This observational, longitudinal, applied study was conducted at the Contact Center of a diagnostic medicine company from October to December 2025. Interactions were managed in Genesys Cloud CX (GC2 license) using speech analytics for transcription, semantic classification, and sentiment scoring. CSAT was implemented on the WhatsApp channel, and all WhatsApp interactions with complete transcripts underwent automated curation and were classified as positive, negative, or neutral. Via API integration, a Python algorithm classified overall sentiment, primary emotions, and calculated the NSS. Results were consolidated in a BI dashboard with daily stratification and a calendar heatmap of positive and negative sentiment. All customer comments were anonymized in accordance with Brazil’s LGPD and comparable international data protection regulations. In Genesys Cloud CX, we recorded 51,492 WhatsApp interactions in October and 44,552 in November. CSAT responses collected via the Ouvimos Você platform yielded 748 comments in October (623 positive) and 455 in November (405 positive). Structured interventions were implemented across four main domains. In Quality and Governance, actions included redesigning the quality monitoring model, adjusting monitoring volume, and standardizing complaint handling. In Workforce Management and Capacity Planning, measures included replacing employees with negative experience impact, reinstating 22 staff members previously on medical leave, reallocating agents from voice to WhatsApp, resizing teams by business group, and extending Contact Center operating hours. Engagement and Culture initiatives comprised recognition programs, institutional campaigns, and celebratory actions. In Training and Technology, continuous technical and behavioral training was conducted, Genesys GC2 modules were reviewed and optimized, CSAT was deployed on WhatsApp, and the operational mode “Central Elite Activated” was adopted. A progressive improvement in NSS was observed. In October 2025, NSS was 78.25%, reflecting an initially challenging scenario. In November, following early interventions, NSS increased to 87.73% (Figure 1). By December, after consolidation of actions, NSS reached 93.33%, classified as “Very Good,” with a predominance of positive interactions. The integration of AI-based sentiment analysis with structured interventions effectively improved customer experience in the Contact Center. The consistent increase in NSS demonstrates that active listening, combined with operational, and workforce adjustments, enables more precise, timely, and sustainable decision-making. These findings underscore the strategic value of emotional indicators for managing customer experience in healthcare services.