Unique user-generated sets like these provided examples of how users paired and combined clothing items that were aggregated on Polyvore from across the web. Studying tens of thousands of sets helped our Product team understand how the various personas on Polyvore liked to define their style. 

Polyvore's extensive Product Catalog contained clothing, shoes, accessories, and beauty products from all over the web,  so clearly, classifying and categorizing the data from these items was critical in building an optimal search product with an enjoyable user interface. The Product team found that among the many attributes that were gathered about each item, the most important ones for users were the item's color, brand name, and the price point. These factors also turned out to be popular keywords used in searches. 

The first launched version of the Taste Feed could determine a user's unique style based on the items they picked from a feed and then calculated recommended clothing items in real time.

Studies suggested that a clean Auto-suggest interface at the search bar could increased successful search result rates by 10x, thereby increasing the overall CTR on shop pages.

While some users liked browsing feeds and digests of fashion trends, other users were on the hunt for a specific item. For these users, it was important to offer plenty of filters and surface multiple options in one elegant view, for a more specifications-driven search experience.