Clever-media: Online store personalization achieves 30% revenue growth with Retail Rocket’s AI platform

Children’s literature is a very delicate retail segment because books can affect the character of the baby. In addition, each child is unique, and their likings are constantly changing. To provide a quality service, the retailer needs to analyze the customer’s behaviour and interests in real-time. How to carry it out in e-commerce? Let’s analyze, here below, Clever-media online store’s personalization and how to achieve revenue growth of 30%.

Clever-media is a children publishing company founded in 2010 that offers a wide assortment of around one thousand books for children and teenagers. Besides its online shop, their books can be found in bookstore chains throughout the country. In 2018, the publishing brand entered the North American market with the opening of a branch office in New York. 

In order to reduce customer search time for a particular book and to offer additional items that may be of interest, Retail Rocket product recommendation blocks were introduced. The value of the implemented Artificial Intelligence platform is not only in a set of self-learning algorithms that allow creating a personalized and unique shopping experience but also in the Growth Hacking team, which iteratively improves each page by choosing the most effective configuration of the recommendations. In this case study, we will talk about fine-tuning algorithms on the home and product pages of Clever-media.ru.

Home page

On a website, the visitor journey largely depends on the first impression generated on the home page, which is where the importance of arousing a user’s interest with its content lies. The best way to achieve it, showing a display on this page with the best offers is the first step. 

However, there is no general and ideal configuration for all online stores. The different algorithms show various results depending on the product segment and the retailer’s strategy. For this reason and to define the optimal configuration of the personalized recommendations for the Clever-media online shop, we carried out several tests. 

Case 1. Testing the effectiveness of product recommendations on the Home page

As part of the recommendation system’s optimization on Clever-media.ru online store, we conducted a study to examine the effectiveness of different algorithms on the recommendation block for the home page. For this purpose, Retail Rocket used the A/B test mechanism.

Firstly, we divided the total of visitors of Clever-media’s online shop into four segments:

The first segment was shown a block with the website best-sellers:

 

 

The second segment achieved the best results of the test with the implementation of personal product recommendations based on the real-time interests and the behavior of each user.

 

 

The third segment was shown a block with the best-sellers within the user’s categories of interest

 

 

The fourth segment, as a control group, was shown no recommendations.

Results

Segment	Effect in Conversion Rate	Effect in Average order value	Revenue estimation
Bestsellers	+ 31.75%	-3.87%	+ 26.65%
Personalized Bestsellers	+ 26.83%	+ 2.61%	+ 30.14%
Bestsellers within the user’s categories of interest	+ 25.56%	-10.56%	+ 12.29%
Control group	–	–	–

 

Results

According to the A/B test, the implementation of “personal interest-based bestsellers” on the recommendation block for Clever-media.ru’s Home page increased conversion by 26.8% with a statistical significance of 96.4%. This, combined with an average order value growth of 2.6% gave a projected revenue growth of 30%.

Case 2. Testing the effectiveness of product recommendations

In the second stage of the process, we added one more recommendation block. To identify its effect on metrics, we conducted a performance study by using the mechanics of A/B testing.

To start, we randomly divided the total of website visitors into three segments:

The first segment was shown two blocks: one with personal recommendations and the other one below with personal bestsellers:

 

 

The second segment was also shown two blocks, but then in reverse order to the first one. This time personal bestsellers were placed above, and personal product recommendations were below.

 

The third segment was shown a single block with personal bestsellers, acting as a control group as it was the one with the best results in the previous test.

 

 

 

Results

Segment	Effect in Conversion Rate	Effect in Average order value	Revenue estimation
Personalized recommendations and personalized bestsellers	+ 1.79%	+ 5.02%	+ 6.90%
Personalized bestsellers and personalized recommendations	-6.32%	-4.77%	-10.79%
Control group	–	–	–

 

Results

According to this second test, by using two product blocks with “Personal recommendations” and “personal bestsellers” on the homepage of the Clever-media.ru boosts conversion by 1.8%. This, combined with a 5% increase in the average order value, gave a projected revenue growth of 6.9%.

 

Product page

This page provides the retailer with an excellent opportunity to describe the features and benefits of the product to push the buyer to add it to the cart. However, the selected item may, for some reason, not fit the user’s interests. To prevent the visitor from leaving the website without purchasing, to offer alternative products is the best option. That benefits both the online shop, which does not lose a customer, and the visitor who finds what they are looking for.

Once the visitor finds what they want, showing related products to complete the cart is very efficient in increasing the average order value. The Retail Rocket Artificial Intelligence platform can achieve this by implementing unique personal recommendations based on each user’s interests.

 

Case 3. Testing the effectiveness of product recommendations

To choose the most effective fine-tune of recommendation blocks on the product page, we conducted a performance study using the A / B testing mechanics.

We randomly divide the total of website’s visitors into three segments:

• The first segment was shown similar products.
• The second segment was shown personal related products.
• The third segment acted as a control group, recommendations were not shown.

 

Results

Segment	Effect in Conversion Rate	Effect in Average order value	Revenue estimation
Related Products	+ 7.34%	+ 0.79%	+ 8.19%
Personalized Related Products	+ 20.80%	+ 4.80%	+ 26.60%
Control group	–	–	–

 

Results

According to the A/B test, the recommendation block with “Personal Related Products” on the Clever-media.ru’s product page increases the conversion by 20.8% with a statistical significance of 99.4%. That, combined with a 4.8% average order value, gave a forecasted revenue growth of 26.6%

Clever-media testimonial

“In book retailing, it is essential to take a personalized approach and create a unique shopping experience for each customer. The recommendation blocks allowed us to automatically suggest the best products based on each user’s interests. Thanks, Retail Rocket team, for your cooperation and for allowing us to be closer to our customers! ”

Uskova Maya, Clever-media online store manager