Machine learning is all the rage these days, but it’s more than just a buzzword: Companies that smartly harness algorithmic intelligence in just the right way stand to make some serious money. That’s exactly what Pinterest is going for with its latest acquisition.
Kosei is a 10-person startup specializing in product recommendations and machine learning. It was just snatched up by Pinterest, presumably for the role it can play in the company’s efforts to monetize its image-sharing social network.
Kosei’s team and technology ought to go a long way in helping Pinterest meet its ambitious goal to “build a discovery engine for all objects.” Having mapped over 400 million relationships between products, Kosei is in pretty good position to help Pinterest better understand the desires and intentions of its millions of users.
For Pinterest, this is not just a gee-whiz technological feat, but a crucial part of how the company can make money. Even something as basic as the classic, Amazon-style “if you like that, you’ll like this” recommendation engine can help Pinterest better target advertisements to users. This is especially valuable on Pinterest, where user behavior says much more about potential purchasing decisions than it does on other social networks. On Facebook or Twitter, sharing a post can imply virtually anything, from “This is funny” or “What an outrage” to “Look how cute these pugs are” or “This long read on the weird racial politics of online dating sure is thought-provoking.”
On Pinterest, life is simpler. In most cases, pinning an item means something more along the lines of “I want to make one of these” or “Note to self: Buy this”–it’s an indication of a user’s taste and personal style. Monetizing that sort of behavior is a much more straightforward task, in theory.
Whatever money-making potential exists here is likely to be super-charged by a little data science and machine learning. At Pinterest, this sort of technology not only goes into recommending products, but also automated image recognition, optimizing ad delivery, and building out a broader machine learning system for making sense of every cupcake recipe and vintage dress you post.