Pins That Predict: A Look at Pinterest's Content Recommendation Engine


Allene Yue

Whether you still use Pinterest or no longer do, I think we can all agree that their content recommendations are absolutely spot on. But how does their algorithm work, and why is it even important in the first place? Pinterest’s mission as a company has always been to deliver the right content to the right people — meaning that a strong content recommendation system is essential to the success of their brand.  But this is not to say that content recommendation models aren’t useful for B2C companies selling services or tangible products either. In fact, it’s just as important.

Let’s go through an example to understand how Pinterest manages to get its users scrolling for hours on hours with no end.

Meet Laura Wong

Laura is a 24-year old woman who is in the process of fitting out the dream cafe she plans to open. Like anyone in her shoes, she needs to find inspiration, and for her, the first place she looks is Pinterest.

Immediately, she makes a quick search. And this pops up.

Let’s say she scrolls for a bit, but she doesn’t really like the stiff layout of the first few pictures and wants something a bit brighter with more color. Suddenly, she sees this — and she’s in love.

She pins it to her board and scrolls down to look for inspiration similar to this one.

Let’s say she saves the 4 posts below.

And now, Pinterest’s algorithm can really get to work. Based on patterns in Laura’s searching, Pinterest may be able to recognize some trends in what she’s looking for. For example, Behance seems to be a platform Laura takes a lot of interest in. As a result, Pinterest might start recommending even more content from Behance.

The majority of these pinned posts are also very Asian-inspired or come from bakeries, coffee shops, and restaurants in the Middle East. Pinterest might take the location or language into account when deciding what else to recommend.

These inspiration pictures also seem to incorporate more neutral and brightly-colored woods, along with a bit of greenery. The keywords embedded in these images will likely be pretty similar because of this. Pinterest will then hone in on words like “neutral,” “modern,” “Asian,” “light,” “bakery,” “tea,” or “coffee shop” in her new recommendations.

But Pinterest doesn’t just store and use this information on their main app. Any new knowledge they gain about user preferences, interests, and likes will be used to keep engaging their users. They’ll use it to send them emails about new content recommendations or board ideas perfectly catered to their personal preferences and up-to-date interests.


But for a company selling a product or service, what good is a content recommendation engine to you? Well, take companies like IKEA, Home Depot, Fiverr, or Glossier. What these companies all have in common is that they use blog articles and written/video content to enhance their value to customers. IKEA, for example, has a blog dedicated to maintenance tips, project ideas, and etc. And in order for this content to reach the right audience, they need to utilize a content recommendation system that will:

  1. Collect and store information about customer preferences, interests, and needs based on customer behavior
  2. Deliver content catered to each customer based on the collected data

And any company could do this with the same success as a brand like Pinterest, so long as they have the right content recommendation engine for it.

Let me tell you a story about customer experience ...

We run a podcast interviewing best in class CX teams. We get to know how they are helping their businesses win.

We'd like to share some stories of what good CX looks like, and what, uhhh! - less than good looks like. Are you in?