If you have ever used Amazon, you certainly have noticed how accurate (and somewhat scary), their recommendations are. Based on your past purchases, ratings, views, searches, and other data, Amazon uses a large data pool of user shopping behavior to offer recommendations by comparing the history of other users and find what patterns resemble yours.

While browsing through the site for some books, I noticed that they present recommendations many different ways. Whether this is just a usability test or intentionally, I’m not sure. The following is a look at some of the various ways Amazon recommends books and the pros and cons of each.

Items you Own

Found in your personalized Amazon section, the list uses your past purchases for their recommendations. The look and feel is similar to Amazon’s site when searching and looking for a product. The difference is the ability to add the product to the cart or your wish list directly and on a full screen, I’m only able to see a little more than 3 recommendations while the normal search shows as much as 5.

Pros:

• Tells you why it is recommended (past purchases)
• Ability to tell Amazon you already own it, are not interested in it, and the ability to rate it
• Able to add to the shopping cart or wish list directly
• Able to sort by All, New Releases, or Coming Soon

Cons:

• Does not say when you can get it by if you order by what time
• Does not say if it is eligible for free shipping
• Only shows 3 recommendations on my maximized browser

Today

Amazon provides a daily sample of recommended items on your personalized page. Recommendations are provided in a horizontal layout where you can flip through pages with left and right arrows. Cloud tags shown below allow you to filter the results. Scrolling over the product produces a small pop-up with product information.

Pros:

• Can rollover to get more information if you are interested, including why it was recommended
• Can filter results by selecting a category in the cloud tag
• Able to add to the shopping cart or wish list directly
• Shows up to 5 products and can easily view other pages

Cons:

• Only shows a sample and not all recommendations (although there is a link to it)
• Have to rollover to get information

Customers with Similar Searches Purchased

This is one of the newer ones that I had not seen before. Products are shown in perspective where one can rotate the products with the closest ones appearing larger. Products can be selected wheather it is in the back or front, even if it is slightly covered by another product. A reflection of the products gives the feature a “Web 2.0″ look and feel.

Pros:

• Able to see and select many products in a small section
• Visually pleasing and provides a fun interaction experience

Cons:

• No product information

Recommended for You

This is found on Amazon’s front page. Directly below it is a “More to Explore” section which gives recommendations based on something you looked at previously. You may also find other similar looking sections based on your browsing history or what customers buy after viewing a certain item.

Pros:

• Able to find out why it was recommended
• Shows percentage of Users who buy a product you viewed and the percentage they buy an alternative

Cons:

• Does not show any product details per section
• Only shows 3 products

Selected for You

This is currently found on Amazon’s front page, towards the bottom of the screen. They only show you one product and a short description but on the side are 4 other “ideas”, including the currently selected one. Clicking on any of the other 3 recommendations will change the picture and description.

Pros:

• Very little screen real estate used
• Interactive interaction experience

Cons:

• Does not say why it is recommended
• Very little detail such as price and rating
• Requires as much as two clicks to get to product detail page

Final Word
With so many ways to recommend products, which one is the best? Only Amazon knows by looking at log data and analyzing customer behavior. Each method has different pros and cons. I personally like the Today’s Recommendation section because it provides me with more choices, options, and detailed information when I want it, although I have never used it before today. In a limited space, I enjoyed Amazon’s “Customers with Similar Searches Purchased”, although more for fun, instead of interest in purchasing. Surprisingly, the one I find myself clicking the most is the one where they give percentages of what other customers purchase based on past behavior. Instead of giving different variations of recommendations, Amazon should improve their algorithm to show their recommendations based on what type the User uses the most.