Pinterest has been developing new machine learning models to improve its home feed, under the banner of “Pinnability”.
“The more people Pin, the better Pinterest can get for each person, which puts us in a unique position to serve up inspiration as a recommendation engine on an ongoing basis,” a spokesperson for Pinterest tells WebProNews. “Home feed is the most heavily user engaged of our products, and is a key way Pinners discover new creative ideas. But with more than 30 billion Pins in the system, we developed smart feed last August to help surface quality, personalized Pins quickly. Pinnability powers smart feed, and applies advanced machine learning models to predict how likely a Pinner will interact with a Pin.”
“For example, if a Pinner follows a dinner recipes group board that has multiple contributors, but isn’t interested in meat recipes, Pinnability picks the most personally relevant recipes for their home feed while de-prioritizing ones that aren’t as relevant (e.g. recipes with meat),” the spokesperson explains. “Our data set is unique in that it contains abundant human-curated content, so that Pin, board and user dynamics provide informative features for accurate Pinnability prediction.
Pinterest says it has seen significant boosts in engagement since launching Pinnability, which has received continual improvements. It’s seen a 20% increase in people who repin from home feed, as well as “significant gains” in total repins and clickthroughs.
The company has a detailed look at Pinnability in a new post on its engineering blog.
Pinterest announced in January that it would start showing Promoted Pins in its home feed. This followed the news that the company had acquired Kosei, which includes, in Pinterest’s words, “some of the best minds in machine learning, data science and recommendation engines, who’ve created a unique technology stack that drives commerce by making highly personalized and powerful product recommendations.”
“With these leaders in machine learning and recommendation systems, we can move faster in discovery and monetization, and building an infrastructure to help brands better understand customer intention and deliver the best content to Pinners,” said Michael Lopp, head of Pinterest engineering, at the time.
The company says Pinnability continues to be a core project in building its discovery engine and that it’s begun to expand the use of its Pinnability models to help improve other products outside of the home feed.
Pinterest has also been hard at work improving Promoted Pins. According to a recent report, it’s about to start offering more precision in its targeting. More on that here.
Image via Pinterest