Top e-commerce recommender systems

Quickly compare e-commerce recommender systems

In e-commerce, a recommender system or engine helps you personalize your visitors’ shopping experiences to drive more conversions. In simple terms, the system serves up recommended products, often based on user demographics (gender, age group, income level, interests), and behaviors (such as previous product purchases, campaign engagements like clicks, and time spent on certain pages). Amazon “wrote the book” on e-commerce recommendation engines and businesses of all sizes are quickly starting to follow suit. The benefits are compelling: about 35% of Amazon’s revenues can be attributed to product recommendations.

How do e-commerce recommender systems work?

E-commerce recommender systems use algorithms to generate product recommendations. “Collaborative filtering” is the algorithmic technique popularized by Amazon and Netflix. It uses aggregates customer details, ratings and reviews to determine recommendations. So called “contextual bandit” algorithms use known visitor data to recommend the best converting offers and promotions before the visitor has made their interests explicit. Text analysis algorithms use Natural Language Processing (NLP) to help you determine which keywords you should use in your content to offer a personalized experience for each visitor.

In practice, these algorithms are typically used to populate the “You might also like” section of e-commerce sites with products that speak to a person’s interest and purchasing patterns. They can also be used to send very targeted email campaigns with personalized product suggestions or to personalize search results.

The tech-speak may sound intimidating, but the good news is: you don’t need to know the details. There are several tools you can use to implement product recommendations with ease. What you will need is data and some time to set things up.

How can you do this yourself?

We made a list of top e-commerce recommender systems to help you review your options.

We looked at:

    • Implementation time and requirements
    • Available features
    • Price
  • Level of support