Saks Fifth Avenue took conversion from 1.5% to 4% with real-time personalisation on SurrealDB
Saks Fifth Avenue is a luxury retailer with $6 billion in annual revenue and 5 million customers. Its product recommendations, 45 million queries a month, run on SurrealDB's vector search and graph.
Challenge
Saks serves 5 million customers and answers 45 million product-recommendation queries a month. That scale needs a database that answers personalised questions in milliseconds, at every hour of every peak shopping period.
Despite heavy traffic, conversion sat at 1.5%, below the benchmarks for luxury retail. Generic recommendations left customers disengaged, and the fix was personalised suggestions delivered in real time.
The data to personalise with existed but was scattered. Preferences, purchase history, browsing patterns and product relationships sat in siloed systems, so a unified view of any one customer was not possible.
Solution
Saks implemented SurrealDB's vector search for recommendations, processing 45 million queries a month with sub-100 millisecond response times. Similar products and similar customers are found by meaning rather than by category alone.
The graph maps 5 million customers to their purchases, browsing and preferences. Customer data from every touchpoint is consolidated into one platform, so the silos are gone and the picture of each customer is whole.
Large language models work with the vector search to deliver contextually aware recommendations that adapt to a customer's behaviour as it happens.
Results
45 million queries a month
For 5 million customers, at consistent sub-100 millisecond response times.
Conversion from 1.5% to 4%
A 167% lift in conversion, and millions in additional annual revenue.
Repeat purchases up 30%
Across the 5 million customer base.
99.99% uptime
Held through peak shopping periods.
