FashionPrimary Source Datasets
Fortune 500 fashion brand
Predicting style preference without purchase history.
The Challenge
A global luxury fashion retailer faced a critical problem. In-store, they knew sales were happening, but they had no visibility into who was making those purchases. Online-first competitors were outpacing them in personalization. New visitors arrived with no purchase history. Loyalty program data was thin. The gap between what they could infer about a customer and what competitors could show them in the first session was widening fast. For traditional retailers competing in an AI-first world, this is the structural disadvantage: your data is fragmented, opaque, and years behind. You cannot build that history overnight. You cannot afford to wait either.
The Solution
The company’s research unit partnered with OpenDataLabs to test a hypothesis: could listening behaviour and other social signals predict fashion preference with enough accuracy to personalize the experience for new users on day one?
We sourced first-party datasets from tens of thousands of consenting contributors. The data linked listening behaviour with fashion preference indicators across multiple domains, ground truth data sourced directly from real users. The research unit built and validated the model internally, testing predictions against baseline demographic and behavioural inference methods.
The Result
The model successfully predicted user style preferences more accurately than baseline methods. For the first time, the company could show a new customer a personalized experience on their first visit. No purchase history required. No waiting for 50 clicks to generate enough signal.
Why It Matters
If you are a luxury or traditional retailer competing in an AI-first world, you are at a structural disadvantage. Your customers do not live in your loyalty program. Your in-store transactions are opaque. Your data is fragmented across channels and systems. Meanwhile, online-native competitors train their models on years of user behaviour.
Ground truth data sourced directly from users, with their permission, lets you leapfrog years of data accumulation. On day one, a new customer sees recommendations shaped by what you already know about their taste. No loyalty program required. No waiting for signal. This is how traditional retailers compete in the AI era.