We’re used to it by now. Netflix recommends movies and shows based on our watch history. Of course, there’s Amazon, who has been giving personalized product recommendations for over 20 years! By now, product recommendation has become so ubiquitous to shoppers that it’s only when product recommendation goes wrong that people take notice. And these recommendations would not be possible without sophisticated AI that parses the mountains of data needed to offer the right product at the right time. Here are 3 ways Amazon uses AI to improve product recommendations, and how AI makes it easier than you think!
Amazon practically invented the concept of giving personalized product recommendations after online purchases, using an algorithm they call “item-based collaborative filtering.” This algorithm makes the homepage of each of its many millions of customers unique, based on their interests and previous purchasing history. The Semantics Scholar article, “Two Decades of Recommender Systems at Amazon.com” puts it best:
Amazon.com has been building a store for every customer. Each person who comes to Amazon.com sees it differently, because it’s individually personalized based on their interests. It’s as if you walked into a store and the shelves started rearranging themselves, with what you might want moving to the front, and what you’re unlikely to be interested in shuffling further away.
Amazon doesn’t only use the purchase data of each of their customers, they also utilize the purchase histories of other people that purchased the same product, giving “frequently bought together” information on their product listings. Furthermore, they factor in customer feedback and ratings. How? By offering recommendations that match a customer’s interests as well as reported customer satisfaction, price and quality level.
Amazon continues to improve their collaborative filtering by connecting purchase history with browsing data. If, for instance, a customer purchased socks, Amazon may not suggest just socks in the future. Instead, their algorithm may look at an individual’s browsing history, see they watch a superhero movies on Prime, and in turn, recommend Marvel brand shirts. To Amazon, interpreting massive amounts of divergent data in real-time is key, and this recommendation engine is responsible for a whopping 35% of their total revenue.
Amazon Alexa and Go
Amazon’s predictive algorithm goes far beyond their website. They now possess a number of alternate purchasing options — such as through Amazon’s voice assistant, Alexa, or their new physical stores, Amazon Go — that make it easier to create a seamless shopping experiences for their customers. Through products like Amazon Echo, Amazon makes it easier for customers to purchase products, but also gives itself other data points, such as what music they are listening to. These data points make it easier to give more holistic suggestions.
Then there is Amazon Go, the company’s foray into physical stores, which we wrote about here. These stores automatically keep track of every customer purchase better than any store currently in existence. As Forbes states, “Data from customers’ smartphone cameras tracks shopping activities and not only helps Amazon Go, but can also be shared with the machine learning team for continued development.”
The takeaway? No matter where they collect their data, Amazon creates as many touchpoints as possible to better understand customers and create fully holistic views of their behavior. Take this example from Forbes “A customer can visit the Amazon Go store to get a few items for dinner, ask Alexa to look up a recipe and the product recommendation engine can determine that the customer likely needs to purchase a certain type of sauce pan.” This integration allows Amazon to create an extremely detailed profile of every one of their customers, and provide multiple ways of delivering exactly what they need through whatever method the person desires.
Introducing Amazon Personalize
Amazon has also gotten into the business of sharing their personalization and recommendation technology with other companies through Amazon Personalize, a machine learning service. The purpose of this service is to “overcome common problems when creating custom recommendations — such as new users with no data, popularity biases, and evolving intent of users — to deliver high-quality recommendations that respond to specific needs, preferences, and behavior of your users.” Already, major companies such as Domino’s, Subway, and Yamaha are partnering with Amazon, recognizing the importance of AI to interpolate customer needs.
Ultimately, you don’t need to be an industry giant like Amazon to handle the enormous amount of data and perfect recommendations. AI technology exists to make this seemingly impossible task easy. Lineate can help personalize product recommendations with its data-driven recommendation engine for companies of all sizes. Contact us for a free evaluation.