The situation
On a marketplace of one-off luxury items, most shoppers come looking for one thing. If they don't find it, or find it and leave, the visit ends there.
The problem
The business wanted more of its visitors to discover other items they might love, without making the site harder to use. Recommending well is tricky when every item is unique and is often sold only once, because there's little history on any single product.
What we did
The team described each item by what it is: the kind of product, brand, price range and other attributes. They then built a system that suggests items with similar or complementary characteristics, so a shopper looking at one piece sees others that fit their taste.
Because the suggestions are based on the items themselves, they work for brand-new listings too, which is exactly what a marketplace of one-offs needs.
The outcome
Shoppers found more items they liked, and revenue rose by 4% through additional purchases from the same visits.
What it meant for the business
A modest, well-placed feature produced a lasting uplift, and the recommendations improved the experience for customers rather than interrupting it.
Could this work for you?
If customers often buy one thing and leave, the answer to "what else might they like?" is probably already in your product data.
Related services: Predictive modelling and forecasting, Customer 360
This project was led by our founder, Achyutanand Jha, in an earlier role. The client is not named, and figures are those reported at the time.