Most people know about Amazon product recommendations for their homepage experience where the recommendations are personalized based on my browsing history and past purchases. We all know that this is a little strange when they continue to show you stuff that you have already purchased or are no longer interested in.
To be sure, Amazon has forgotten more about personalization and product recommendations than most companies will ever know. To get a lesson in what really matters, search for or navigate to a product you have never shown interest in before and see what happens. My product was a pressure washer. When I navigated to a particular product page the recommendations became very contextual (i.e., they were based on that product instead of my past purchases). Amazon chooses three options here: frequently bought together, customers who bought this item also bought, and compare to similar items.
First, these options help me complete my purchase and second, they make sure that I get an item I will be happy with and they also ensure I will get all of the accessories I need to be successful with my project. Amazon knows that there is nothing worse than getting your new item and not being able to use it. By the way, I am getting the sense that I really need a brass hose connector, some hose washers and some cleaning concentrate to go with my purchase. Since this is a new area for me, the option for foam lance is something obviously I need to explore.
So, for those of you who thought that relevant personalized content was always about user attributes and past purchases take a look at how the pros use context to deliver a scalable, relevant user experience for millions of products on their site.