Key results
The company
Grupo Casas Bahia
grupocasasbahia.com.brOmnichannel retailer of electronics, furniture, and appliances in Brazil.
Result highlights
- ~R$480,000 annual operational savings
- 4,000+ person-hours saved annually
- 9+ hours saved per 1,000 comments
- Monthly review processing increased from 1,500 to 33,500
- 90% accuracy in customer journey identification
- 14x gain in analytical efficiency
The story
One of Brazil’s largest omnichannel retailers serves over 100 million customers through more than 1,000 stores and a massive national logistics network.
Processing thousands of daily reviews from diverse channels created a bottleneck, with staff requiring an hour to manually classify just 100 comments. This labor-intensive process forced reliance on small data samples that left significant gaps in understanding customer dissatisfaction.
The team consolidated feedback from internal and external sources into a secure lakehouse to create a unified view of the customer voice. They deployed a custom prompt for Meta’s Llama 3.3 model to automate the classification of comments by journey stage and specific problem type. The resulting structured data feeds daily dashboards that enable product and UX teams to prioritize fixes based on real-time insights.
Quotes
“Databricks is a revolutionary facilitator for our AI strategy. The platform lets us innovate, collaborate, and deliver insights at scale — automating what used to be slow, manual work and giving every area of Casas Bahia a real-time, 360-degree view of the customer’s voice.”