Impact of Recommendations on Customer Engagement

Why should you power your eCommerce site with intelligent recommendations?

By Gopi Joshi
Published on November 22, 2020

In the past couple of years, eCommerce has taken huge strides to provide a better and personalized customer experience. Recommendation sections on eCommerce site have taken the role of the traditional salesperson in brick and mortar system. With the advancement in artificial intelligence, recommendation engines are constantly evolving to understand the user better and provide meaningful recommendations. This not only improves the purchase experience but also increases revenue for eCommerce stores.

Several studies and surveys show the importance and impact of recommendations on customer engagement and sales.

  • According to a study(1), online retail browsers who engaged with a recommended product had a 70% higher conversion rate during that session.
  • According to a survey(2), 38% of US digital shoppers said they would stop shopping at a retailer that made poor product recommendations.

With Intelligent Recommendations Engine, eCommerce stores can enhance the customer experience, user engagement, and top-line revenue with new and repeat purchases.

In the eCommerce journey, users create so many events and generate so much data, which if used wisely, can elevate their experience and engagement. Recommendation systems use all the data produced by the user to provide engaging and relevant content to the user. By using catalog data and behavioral data, recommendation engines suggest the right product to the right user at the right time and delight users by understanding their needs and likings.


Netflix and Amazon, both industry leaders, are great examples to illustrate the power of recommendations. Amazon has accounted 35% of its revenue to its powerful recommendations, while Netflix believes if it had not implemented a personalized recommendation engine, it would be losing $1 billion or more due to cancelled subscriptions.

How Amazon uses recommendation engine?

The recommendation system contributes heavily to Amazon’ s success. According to research by McKinsey, almost 35% of its sales come from recommendations. It is believed that Amazon’ s growth in the recent times has been majorly due to the way they integrated different recommendations throughout the consumer’ s buying lifecycle. With a data-driven approach, Amazon has successfully captured a big chunk of market share.

All the recommendations can be divided majorly into two segments.

  • Onsite recommendations.
  • Email recommendations.

According to Sucharita Mulpuru, a Forrester Analyst, the conversion rate of onsite recommendation could be as high as 60%.(6)


The following are some of the onsite recommendations.

  • Recommendation for you – Personalized: Amazon personalizes recommendations based on the categories that the user has been browsing and might be interested in buying.
  • Frequently Bought together: This recommendation aims at increasing the average order value based on upsell and cross-sells.

How Netflix uses recommendation engine?

In the 1st Quarter of the year 2020 alone, Netflix has added 15.8 million more subscribers(5), which attributes to 22% growth, registering quarterly revenue of $5.77 billion. Netflix now has over 182 million subscribers worldwide.

Netflix has changed its model from renting/selling DVDs to content streaming. While changing the model, it ran a huge contest from 2006 to 2009, asking different teams to design an algorithm that can improve the performance of its famous in-house recommender system ‘ Cinematch’ by 10%. The team providing the best improvements was to be awarded $1 million. The dataset given to the teams consisted of 100,480,507 user ratings that 480,189 users gave to 17,770 movies. In 2009, the prize was awarded to a team named BellKor’ s Pragmatic Chaos.

After implementing various recommendation systems, the overall user engagement rate of Netflix increased to a great extent. It also increased streaming hours and lowered cancellation rates.

Netflix believes that personalization and recommendations have saved more than $1 billion per year for them. 75% of the content that people watch on Netflix is provided through recommendations.



Below are a few key technology trends for recommendation engines.


Content-based Filtering In this method, the recommendation system tries to find products that are similar in terms of content. When the user is looking at some product, this type of recommendation shows other products that are similar in terms of various attributes like size, color, brand, etc. This type of recommendation system heavily relies on catalog data that encompasses various product details.

Collaborative Filtering This type of recommendation system tries to find a similar user based on various user data and tries to recommend products based on the behavior or actions of other similar users. Such systems consider things like purchase history, profile data, reviews, and ratings, browsing behavior, etc. to find similar users.

Hybrid Many recommendations take a hybrid approach with a combination of content-based and collaborative filtering. Many recommendations use demography, gender, profile, etc. for segmentation.


Different types of recommendations are seen on eCommerce sites on the pages throughout the customer’ s journey to purchase. Some of the recommendation types are as follows:

View – View

This type of recommendation suggests the products that were viewed/browsed together.

Affinity to Recently Viewed

This type of recommendation shows a list of products based on browsing history. It considers the products in browsing history and information of various dimensions of products in the same category. Recommendations are given based on similarity among various attributes.

Bought – Bought

This type of recommendation gives users a list of products bought together.


This type of recommendation shows top trending products. It is generally used in the sale or festive season. Based on customers’ actions like browse, add to cart, add to wish list, purchase, etc. the trending items are decided. It keeps changing very fast based on behavior data and inventory as festive seasons see heavy buying.

Frequently Bought Together

This type of recommendation shows a list of products that are generally bought with this product in the same order. This recommendation aims at increasing the average order value based on upsell and cross-sells.

Related Products

This type of recommendation shows products that are related to the product getting browsed. This section either uses collaborative filtering, content-based filtering, or any hybrid approach in determining the relatedness of the products.

Best Sellers

This type of recommendation shows best-selling products. Depending on where this section appears on the site, companies display best sellers for a specific category, best sellers for specific brands, and best seller across various categories.

Personalized Recommendation

Many sites also show personalized recommendations based on user’ s browsing data, purchase history, profile information, and demographics. Personalized recommendations are useful in giving the right content to the right user at the right time. Because personalization focuses a lot on finding the most relevant products for the user, it may significantly improve user engagement and satisfaction.

Recently Viewed

This type of recommendation shows all the products that were recently viewed by the customer. It helps the user to continue browsing from the previous session. This way the recommendation engine can ensure that the user does not miss out on any product that they were interested in previously.


ARC’ s Recommendation xperience Engine

Recommendation section on a commerce site gives a window to the user to consider something more. Nowadays, a lot of user data is available to generate useful insights. Many impactful mining techniques and effective recommendations help retailers to continuously engage, enhance customer experience and loyalty.


ARC’ s Recommendation xperience Engine provides a wide range of recommendations for each page of an eCommerce site. ARC’ s data-driven approach helps in increasing user engagement in each stage of the user journey. The data ingestion module helps businesses utilize the latest data and capture latest trends to generate intelligent and most relevant recommendations. ARC’ s built-in analytics allows businesses to measure KPIs associated with different recommendations, enabling retailers to make better decisions for future recommendations.



About the Author


Gopi Joshi

Product Manager managing products like Search and Recommendation. eCommerce enthusiast, helping customers shop better.

Get in touch with us

We are eager to discuss your business needs and answer any questions you may have. Enter your details and we will get back to you shortly.

+1 888-627-7657

Please enter first name
Please enter last name
Please enter company name
Please enter valid email
Please enter phone