Have you ever shopped on the website www.amazon.com, one of the largest online retailers in the world? When you search Amazon for a particular book, the site also displays other books that are frequently bought together with the original book. It will also show you what books other customers bought with this book. Very often, customers find these recommendations very useful and will buy more than just the one book they were looking for. Have you ever wondered how Amazon manages to provide recommendations that are so relevant and hence so useful?
No, Amazon has not employed thousands of avid readers monitoring every visitor on the website so they can provide personalized recommendations as a visitor searches for a book.
Instead Amazon uses statistical algorithms to analyze user information captured on its website enabling it to offer powerful recommendations to its customers that are purely driven.
In other words, the recommendations are not based on someone’s intuition or experience or feeling but is based solely on how other customers have behaved in the past. Behind these offerings is a statistical algorithm, based on conditional probability, which calculates likelihood of 2 or more items being purchased together.
Yet, or probably because of that, the users find these recommendations extremely useful.
If the customer is happy with the recommendations he is more likely to buy more items, increasing Amazon’s bottom line. Customers are also more likely to return to the website if they are happy with the experience leading to an increase in the lifetime value of the customers.
This kind of innovative use of analytics to drive decisions and strategies has led to Amazon being hailed as a global e-commerce powerhouse that did nearly $40 billion in sales last year, dealing in everything from banjo cases to wild boar baby back ribs.