Galeries Lafayette operates 335 stores and a fast-growing e-commerce business that attracts more than 60 million visitors annually. As demand for hyper-personalized shopping experiences increased, the company needed to modernize its search and merchandising capabilities to improve relevance, increase conversions, and reduce operational complexity.
Its legacy search stack struggled with long-tail search accuracy, contextual recommendations, and differentiated customer experiences across a catalog of more than 600,000 owned and marketplace SKUs. Merchandising teams also faced growing manual workload and limited experimentation capabilities. To modernize commerce experiences, Galeries Lafayette partnered with Grid Dynamics and Google Cloud to deploy an AI-powered platform built on Vertex AI Search for Commerce and Grid Dynamics’ Merchandising Experience Platform (MXP).
See the architecture behind 7% revenue growth and 2X faster merchandising
The new platform integrated Vertex AI Search with SAP Commerce Cloud, Mirakl, and MXP to improve search, recommendations, and navigation across web, client, and seller applications. Built on a Google Cloud foundation, the solution introduced automated catalog, pricing, and stock ingestion pipelines alongside flexible merchandising controls that simplified rule management, rollout workflows, and experimentation.
The initiative delivered measurable business impact:
- 7% total revenue increase
- 8% increase in average basket value
- 20% year-over-year increase in online sales during peak retail periods
Following a phased rollout from A/B testing to full deployment, the company saw stronger conversion performance, lower bounce rates, and deeper product engagement across customer journeys.
Galeries Lafayette’s discovery modernization also created a scalable foundation for future agentic commerce experiences, including intelligent shopping assistants, advanced personalization, and “shop the look” capabilities.
Explore the full case study for in-depth Vertex AI Search and MXP architecture and implementation details
You might also like
