Amazons recommendation algorithm

Amazons search engine searches within its product inventory to present you the best possible shopping experience? Every activity from the time you start browsing Amazon is tracked, sent to their database and analyzed to present you with most precise and to the point product recommendations. Recently it became more noticeable to me because of my shopping experience there when I wanted to buy products for making Bath Bombs and had no clue what to buy.

Amazons systems predict users interest and give them recommendations to buy related products. On the web page, the recommendations usually list as

  • Related Items
  • Frequently Bought Together
  • More Items to Consider
  • Inspired by Your Shopping Trends
  • Inspired by Your Browsing History
  • Additional Items to Explore
  • New for You
  • More Top Picks for You
  • Customers Who Bought This Item Also Bought
  • Sponsored Products Related To This Item
  • Customers Viewing This Page May Be Interested In These Sponsored Links
  • Your Recently Viewed Items and Featured Recommendations (Inspired by your browsing history)
  • Bought together
  • You may also like.
  • Product matching your recently viewed items

To view my experience, please click on the video below.

Recommended products seem simple and straightforward feature on any website, but there is a lot of complex calculations and combinations going on behind the scene to present the user with appropriate recommendations.

Each product is mapped by users interest for examples

  • Users shopping cart
  • Users browsing pattern
  • Users purchases
  • Users purchase history
  • Product Keywords
  • Product categories
  • Users wish list
  • Abandoned shopping carts etc.

Amazons product recommendation

Amazon uses data mining method where the software programs analyze users activities on the website to generate data values representing degrees to which specific items relates to one another.

It tracks user behavior and interest by collecting a list of products the user is viewing in any given time frame or session. The program then associates the product with each other and compares or look for associations of product from already collected data and store the items identified as related items. These items are tagged as related from various categories based on users browsing patterns which include users buying them together, Users adding them to same shopping cart, users viewing them in the same browsing sessions.

How it works.

To create the relations between products, the system records interest and likings and then analyzes data collected from

  • Purchases
  • Clicking on items to view the detail page
  • Adding items to the shopping cart.

The details are stored in the table, and correlations are created (e.g., products A and B are similar because a significant number of those who bought A also bought B).

Another way to creating relation is data mining from the viewing history of the user. The items viewed in combination are matched and stored as related items. (E.g., products A and B are similar because a significant number of those who viewed A also viewed B during the same browsing session).

To collect this type of data user specific, or session specific, histories of item details page is analyzed. This method is particularly useful when the product itself is new and not been purchased or rated/reviewed by users. By this method data could be collected for the product, which is not rated or not been purchased yet.

Amazons shopping lists

Amazon lets you store multiple shopping lists

In Summary, all or any of the below could be used to collect data and present the use with most relevant recommendations.

  • Shopping card purchased items
  • Abandoned cart
  • A/B testing for different price for the same product
  • A/B testing for same product in different package or differently bundled discount
  • Data from various shopping list of the same member
  • Wish lists save for later list
  • Site browsing behavior and drop-off.
  • Ratings and reviews
  • Demographic details like your IP address, postal address, etc.
  • Coupons, which you used in the past
  • Email click-through or Search engines search results click-through


Pin It on Pinterest