Polyvore is best known for being that site fashionistas use to quilt together gorgeous collages of fashion items, creating their own magazine-like spreads. But the company has much bigger aspirations; namely it wants to be the portal where fashion and lifestyle shopping begins.

And today, Polyvore is taking its first step in that direction by rolling out personalized recommendations in its users’ feeds. Until now, users only saw content from brands and users they follow, much like on Twitter (pre-promoted posts). The new recommendations will be based on the “Style Graph” Polyvore has been stitching together for users based on who they follow, the sets they make and items they use, and other behavioral data.

From a Polyvore spokesperson:

With the new personalized experience, we’ve built a unique style profile for each user. Our technology looks at billions of data points generated from our community to understand products users have liked, disliked, commented on, or incorporated into an outfit. Our algorithms understand attributes associated with those products such as colors, fabrics, and patterns.

Polyvore has been piloting this, showing users 2.5 times more styles in their feeds, and getting four times more product “likes” from them, the spokesperson said.

Recommendations are nothing new in the world of visual-heavy social networks. Pinterest has been beefing up its search and discovery capabilities for a while now, using a sophisticated recommendation engine to power its new Guided Search, as well as using “Related Pins” to introduce users to content they might enjoy based on their pins and activity.

Beyond higher engagement, as demonstrated by the increase in “likes,” I’d venture to say that this is only the first step in Polyvore’s bigger plan to become the portal from which people shop. A few months ago, Polyvore chief executive Jess Lee said that she definitely wants her company to be a place where shoppers begin a purchase, creating sets around items or styles they’re considering and figuring out what they want to purchase. Polyvore’s items are shoppable — the company gets a fee for facilitating a sale — so finding new ways to get people to shop through its site is in its best interests.

The company also likes to describe itself as a “customized boutique” for each user, pointing to the idea that it could someday be the first stop for anyone doing some online shopping. If Polyvore can recommend items to me based on all the data it has on my fashion taste and if I can purchase any and all items right from the site and browse and stay connected to brands I like, why shouldn’t I make it my first stop every time I want to buy a new outfit?

Although this is Polyvore’s first foray into building out its Style Graph, the company will be unveiling more information about the fancy math under the hood in the coming weeks.

Along with fashion, Polyvore’s catalogue also includes home and beauty items, though fashion remains its biggest, strongest, and most core category.

Style Graph