Personalization of the online store Kcentr turned into 60% revenue growth
Household appliances and electronics are a special segment of online retail. There are practically no spontaneous purchases here, customers take their time and carefully evaluate the products offered, comparing them to each other. Product recommendations help the online store to increase the conversion, revenue and average order value. In this case study, we present the personalization of Kcentr and their revenue growth due to the introduction of product recommendations on the “key” pages of the webshop.
Kcentr is a retail chain selling household appliances and electronics of well-known global manufacturers. To provide professional and personal customer service online, the company uses recommendations on the key pages of the webshop.
Case 1. Testing the recommendations effectiveness on the home page.
One of the advantages of online retail is the ability to show the widest range of products. But here lies the problem: how to help the visitors find exactly what they need in the shortest time? Especially on the home page, where you want to show as much from the product portfolio, promotions and other offers. This is when personalization comes to the rescue, adjusting the webshop in real time to the needs of the client.
A study was conducted on the effectiveness of various product recommendation algorithms on the store’s main page. The effectiveness study was conducted using A / B testing mechanics. All site visitors were randomly divided into four segments:
The first segment showed bestsellers
The second segment showed personalized bestsellers
The third segment showed popular products from categories of interest to the user
Recommendations were not shown to the fourth segment. This was the control group.
Results
Following the outcome of the A / B testing, the following results were obtained:
| Segment |
Conversion increase |
Change in average check |
Revenue (per visitor) increase estimate |
|
Bestsellers |
-7.83% | + 2.72% |
-5.32% |
| Personal sales hits |
-2.16% |
+ 3.33% |
+ 1.10% |
| Most popular products from categories, that visitors are currently interested in |
-1.85% |
+ 8.86% |
+ 6.85% |
| Control group | – | – | – |
Thus, the best result was shown by the algorithm “Popular products from categories of interest to the user.” Despite a slight decrease in conversion, the mechanics showed a significant increase in the average check by 8.86%, which ensures the projected revenue growth of 6.85%.
Case 2. Testing the effectiveness of recommendations on the category page
Kcentr added a product recommendations block on the category page to help the users explore their catalog. When there is a profile of the user’s history and interests available, it is used to personalize these recommendations. If not, this recommendation block takes into account the “wisdom of the crowd”, displaying relevant products based on the preferences of other users with similar profiles and interests.
To find out which algorithm of recommendations will show the greatest effect on the category page, a study was conducted using the mechanics of A / B testing. All site visitors were randomly divided into 4 segments:
The first segment showed the most popular products from the category
The second segment showed sales hits from the category, personalized according to user interests
The third segment showed personal recommendations of products from the category
The fourth segment was the control group: recommendations were not shown to users.
Results
Following the outcome of the A / B testing, the following results were obtained:
| Segment |
Conversion increase |
Change in
average check |
Revenue increase estimate |
| Bestsellers from category |
+ 21.96% |
-14.36% |
+ 4.44% |
| Personalized sales hits from category |
+ 9.83% |
-5.30% |
+ 4.01% |
| Personal recommendations of products from the category |
+ 16.36% |
-7.91% |
+ 7.16% |
| Control group | – | – | – |
According to the results, the use of the mechanics “Personal recommendations of products from a category” in the recommendations block on the category page of the online store Kcentr increases conversion by 16.4% with a statistical significance of 99.7%. In combination with a decrease in the average check of 7.91%, this gives a predicted revenue growth of 7.16%.
Case 3. Testing the effectiveness of recommendations on item cards of the product page
The online store has a significant advantage over the offline point due to the fact that it can provide more information in the product page. But the most interesting is the personal recommendations of similar and related products. In an offline store, this mainly depends on the salesperson and the products he will recommend, what he will offer the client to complement the purchase. While in the webshop the role of the salesperson is played by personal recommendation blocks that adapt to each user visiting the store.
As part of the optimization of the recommender system on the site Kcentr, a study was conducted on the effectiveness of various algorithms in the recommendation block in the store’s product page. The effectiveness study was conducted using A / B testing. All site visitors were randomly divided into 5 segments:
The first segment showed similar products
Related products were shown to the second segment.
The third segment showed two blocks simultaneously: similar products (above) and related products (under the block with similar products)
The fourth segment was shown two blocks simultaneously: related products (above) and similar products (under the unit with related products)
The fifth segment was a control group: recommendations were not shown to users.
Results
Following the outcome of the A / B testing, the following results were obtained:
| Segment |
Conversion increase |
Change in average
check |
Revenue increase estimate |
| Similar products |
-4.95% |
+ 5.59% |
+ 0.36% |
| Related products |
-17.55% |
+ 38.60% |
+ 14.28% |
| Similar products (above) and related products (below) |
-8.71% |
+ 46.76% |
+ 33.98% |
| Related Products (top) and similar products (below) |
+ 16.54% |
+ 37.39% |
+ 60.12% |
| Control group | – | – | – |
According to the test results, the use of the mechanics “Two blocks at the same time: related products (top) and similar products (under the block with related products)” in the product card of the online store Kcentr the conversion increases by 16.5% with a statistical significance of 99.9%. Combined with an impressive 37.39% increase in average check, this provides a projected revenue growth of 60.12%.
Case 4. Testing the effectiveness of recommendations on the product page
After selecting the most effective group of recommendations on the product page, we focused on fine-tuning the recommendations of similar products. The effectiveness study was conducted using A / B testing mechanics. All site visitors were randomly divided into 3 segments:
The first segment showed similar products
Similar products were displayed to the second segment with an emphasis on the products viewed by the user.
Similar products personalized to the third segment were shown.
Results
Following the outcome of the A / B testing, the following results were obtained:
| Segment |
Conversion increase |
Change in average
check |
Revenue increase estimate |
| Similar products | – | – | – |
| Related products with an emphasis on viewed products |
+ 0.41% |
+ 5.31% |
+ 5.74% |
| Personalized similar items |
-0.90% |
+ 4.46% |
+ 3.51% |
According to the test results, the use of mechanics “Similar products with an emphasis on products viewed by the user” in the recommendation block in the product card of the online store Kcentr increases conversion by 0.41% and the average check by 5.31%, which gives the predicted revenue growth by 5.74%.
Comment Kcentr
When we started the work on personalizing the site, we didn’t think that the results could be so interesting. Many thanks to the Retail Rocket team for their approach to work. We are very pleased with the performance that we get through the introduction of site customization. This helps us to embody our motto “Everything is for our own people” and increase the conversion, the average check and the revenue of the online store. ”
Mikhail Gudkov, Head of Internet Marketing Department at Kcentr online store