Thursday, August 16, 2018

Pinterest Improves Related Pin Recommendations, Increasing Engagement and Activity

Pinterest Improves Related Pin Recommendations, Increasing Engagement and Activity

Pinterest has quietly rolled out a new element within its sharing algorithm, which has improved recommended Pin results significantly, and is, according to Pinterest, “the largest application of deep graph embeddings to date.”
As explained on the Pinterest Engineering blog, the new process, called PinSage, uses contextual information from surrounding Pins to provide more accurate recommendations for additional matches, as opposed to using just the image or keywords to highlight such.
“The benefit of our approach is that by borrowing information from nearby nodes/Pins the resulting embedding of a node becomes more accurate and more robust. For example, a bed rail Pin might look like a garden fence, but gates and beds are rarely adjacent in the graph. Our model relies on this graph information to provide the context and allows us to disambiguate Pins that are (visually) similar, but semantically different.”
The explanation is overly technical, but essentially, Pinterest is now showing users better matches in relation to what they actually go looking for, which has lead to improvements in both engagement and time spent.
Here’s an example of the system in action – through various machine learning approaches, different related items are highlighted, comparative to the initial basis Pin. As you can see along the bottom line, PinSage has a better understanding of the broader context, as opposed to merely matching with the visual or likely similar elements.
A graphic looking at how Pinterest's PinSage works
As a result of this, tests have shown that PinSage-based recommendations have lead to around 30% relative improvement in user engagement rates - and importantly, a 25% increase on impressions for the platform’s ‘Shop the Look’ product.
It’s a small update in the broader context, for sure, and there are not any actionable takeaways you can implement as a result. But it is worth noting the advances Pinterest is implementing, and how they relate to more accurate recommendation matches.
This is especially relevant when you also consider that Pinterest, which is focused on driving eCommerce potential, sees significantly higher purchase intent among users than other social platforms.
With Pinners coming to the platform looking to buy, and Pinterest showing them more relevant matches for such, it’s worth paying attention to such developments, and considering where Pinterest might fit within your digital marketing approach.