Personalization is central to Amazon‘s business, and has a direct impact on their bottom line. They’ve personalized over half of the real estate on their home page (as shown by the green boxes above) because they make more money by shortening the distance between you and something that you want.
Amazon is well known for using the shopping patterns of other customers to make contextual recommendations of products related to whatever you’re viewing. This works but they need to do more because their catalog is so large that most people barely scratch the surface. Personalization is key to how they drive discovery, and help users find products that they never knew existed but have to buy on first sight. Diversity of recommendations is important, however Amazon is in the business of moving real world inventory, and margins matter. You can expect their algorithm to bias toward recommending best sellers or poor sellers over everything in between to maximize efficiency.
Personalized movie recommendations make up the majority of Netflix’s movie browsing experience. They’re so important that Netflix famously offered a $1 million prize for the best ratings prediction algorithm to improve their results. Take a look at the section headings in the screenshot above, you’ll see that they recommend fine-grained categories, within which they’ll show popular and personalized titles. This blends categorical relevance with popularity and curation, and gives better results than Amazon in my personal experience.
Netflix’s recommendation algorithm was originally developed for their DVD rental business which has a very different model than their streaming business. Contrasted with Amazon, Netflix pays higher licensing fees when it streams best sellers so you might expect them to have a bias to steer you toward more obscure titles in the corners of their catalog.
Yahoo has recently embraced personalization, and their site learns from your behavior like Amazon and Netflix. If you’re a casual visitor you’ll notice Yahoo’s stories becoming more relevant over time. They also support declarative personalization which allows users to take control of their experience and pick categories they like. This could easily become a chore, but they make it really easy to bootstrap by importing your interests from your Facebook profile. Declaring your interests can actually be fun if it’s done in context with the content you’re consuming, with an immediate reward loop that refreshes with better articles.
You can have great data but you can’t personalize without a site layout that’s easy to dynamically reconfigure. Yahoo uses infinite scrolling on their home page, and shows sponsored content (also personalized) inline with their articles. This layout is easier to personalize and makes it effortless for users to scroll until they find something they like.
Pinterest takes dynamic layout and infinite scrolling to another level, and may be a good design reference if you want to use this on your site. Mashable is another site that is experimenting with these techniques, using a mix of standard and native ad units.
In summary, if you’re looking to personalize your site, start thinking about the data you can use to surface more relevant content, and make sure your layout will support the dynamic decisioning you want to do. If your template is feeling stiff and creaky, it might be time for a tune up. In the meantime our Smart Layers can help.