Case Study: Flaconi achieves real-time personalized recommendations and 200x faster model deployment with Databricks

Personalizing the beauty product shopping experience with Data + AI

Flaconi, a leading European online beauty retailer with over 55,000 products and 2.3 million customers, needed to scale personalization across massive streaming and historical datasets while reducing infrastructure complexity and supporting both analysts and data scientists. Rapid growth, integration challenges, and the desire to experiment with deep learning, NLP and image processing made it difficult to deliver real-time, user-centric recommendations at scale.

Flaconi implemented the Databricks Lakehouse Platform on AWS—using Delta Lake for reliable streaming pipelines, MLflow for model lifecycle management and integrations with SageMaker—to deliver streaming analytics and real-time recommendations. The platform cut infrastructure setup from two days to 15 minutes, sped model deployment 200x, reduced prediction latency from 20 minutes to 300 ms, lowered staff costs by 40%, and is expected to lift net order income by about 5% while increasing cart values and conversion.

Flaconi

Carlotta Schatten

Data Science Team Lead