Default Banner

Pinterest Launches A Shiney New Algorithm Called Pinnability

Pinterest is beginning to poise itself as a search engine, or what they call a “discovery engine”. Taking cues from Facebook, Pinterest is now serving up what they call a “home feed”. This feed is powered by Pinterest’s Smart Feed algorithm that was launched in 2014.

So what is Pinnability? It’s a collection of machine learning models that allows Pinterest to serve a smart feed on a Pinterest user’s homepage.  Pins used to be in chronological order, but now it’s the job of Pinnability to order pins based on a relevance score, or more specifically, a personalized relevance score. Pinterest states that filtering by chronological order isn’t enough to help Pinterest users see pins they really want to see.

Pinterest has a Pinnability diagram on their blog to show the logic of this machine learning model.

Pinterest Home Feed Before Pinability

Pinterest Home Feed After Pinability

Machine learning models have always played a significant role in the development of Pinterest. Pinnability is powered by smart feed, the technology that brings flexibility and control over the composition of the home feed. Pinnability and smart feed work together to estimate the probability of a Pinner interacting with a Pin, with pins receiving high relevance scores shown first. The personal relevance score includes all Pins that users follow as well as Pinterest’s ‘Picked for You‘ recommendations. The algorithm takes into account ‘training factors’ like Pin quality, historical popularity, authenticity, and relevance. The demographics and the activity levels of Pinners are carefully analysed and entered into the mix. And the result is a personal relevance score for each Pin. So in theory, pinner interaction improves because their feed is more interesting and useful due to the personalization.

Per Pinterest’s statement on the subject, “We continue to refine Pinnability and have released several improvements to-date. With each iteration, we’ve observed significant boosts in Pinner engagement, including an increase in the home feed repinner count by more than 20 percent. We’ve also observed significant gains in other metrics including total repins and clickthroughs.” Which means their accuracy of recommendation quality is still a work in progress.

For more information on Pinterest and social media marketing, Contact Position². We love giving free audits!

Team Position2

March 24, 2015

By Team Position2