Log in

BurdaStyle & Retail Rocket: mobile website personalization achieves 27,7% conversion uplift

BurdaStyle & Retail Rocket: mobile website personalization achieves 27,7% conversion uplift

Nowadays, users frequently shop via smartphones. How can we maintain users engagement when users are visiting us via their mobile phones? Do they or do they not perceive desktop and mobile product recommendations equally?

According to Wolfganf report the number of orders via mobile devices has increased by 11% in 2018 and so online stores generate up to 50% of overall revenue from mobile users orders. We cannot ignore this trend, so we have prepared this case to explore recommendations performance on mobile phones by using AB-testing.

About BurdaStyle

BurdaStyle is a specialized well-known media title covering trends in sewing, needlework, creativity, style and fashion. BurdaStyle is a great place to shop for patterns and special goods for sewing and needlework. You can also find sewing tutorials, articles on fashion and tips from well-reputed stylists.

The challenge

BurdaStyle users often shop via their smartphones, so we had to fine-tune mobile version personalization with Retail Rocket algorithms. Therefore, we focused on achieving the following objectives: 

  • Discovering the best recommendations algorithms for key mobile webshop pages;
  • Increasing uplift metrics, especially conversion rate;
  • Comparing recommendations performance for desktop and mobile webshop.

Retail Rocket solution

Some tests were executed on desktop version before we set up mobile recommendations. The results were pretty good but we rejected the idea of implementing desktop winning configuration without changes, even though it seemed to be the fastest way.

The thing is that smartphone and desktop users have different shopping profiles. As Retail Rocket key objective is to increase the online store revenue, we always get into each case individually to find a unique solution. For BurdaStyle the solution was to test brand new personalization algorithms on the mobile version of their webshop.

Home page recommendations efficiency testing: two variations for the start

What can we offer a user on our home page? Some online stores demonstrate a standard product compilation from CRM formed on an abstract parametric basis without taking into account the demands of real users. 

We placed Retail Rocket personal product recommendations on the mobile version of their home page. This algorithm accumulates a great amount of user’s behaviour data, so our auto-generated recommendations correspond to the interests of real users.

We executed an A/B-test to check recommendations performance where users were randomly split into two segments.

Personal product recommendations were shown to the first segment:

Whereas no recommendations were shown to the second segment which was the control group.

Results

The A/B testing showed the following results:

Conclusion

According to the test results, the use of «Personal recommendations» algorithm for BurdaStyle home page increases conversion rate by 24,7% and its average order value by 0,9%. This all together reaches the estimated revenue growth of 25,7%.

Note: The winning configuration for BurdaStyle desktop home page was different, even though it was also based on personal product recommendations. We compared the results of the desktop and mobile versions and learned that we needed to find the best algorithm for each particular page. 

Category page: multivariate testing

We decided to test personal product recommendations for the category page as well and execute an additional experiment. Some users do not accumulate enough data to generate an individual product compilation, so can the temporary placement of popular products recommendations compensate for this issue?

Let’s check it! We executed an A/B-test with three randomly split users segments.

Personal product recommendations from the category were shown to the first segment. If there was not enough data for a particular user, such customer saw no recommendations.

Personal product recommendations from the category were also shown to the second segment. Users without browsing history saw popular products from the category.

No recommendations were shown to the third segment which was the control group.

Results

The A/B testing showed the following results: 

Conclusion

According to the test results, the use of «Personal recommendations» algorithm for BurdaStyle in the category page increases conversion rate by 1,7% and its average order value by 4,5%. This all together reaches the estimated revenue growth of 6,2%.

Despite the average order value uplift, the second segment conversion rate significantly decreased. So we were convinced that personal recommendations algorithm performs more effectively for category page without changes. 

Product page: conversion uplift through alternative and related products recommendations

The product page is unique for each online retailer. We cannot predict beforehand which algorithm fits best the particular store. That is why we always pay special attention for product page personalization.

We have tested five algorithms variations using the A/B-testing approach. All website visitors were randomly split into five segments in real-time:

Alternative products were shown to the first segment: 

 

Related products were shown to the second segment:

Two recommendations blocks were shown to the third segment: alternatives (above) and related products (below alternative products):

Two recommendations blocks were shown to the fourth segment: related products (above) and alternatives (under related products):

 

No recommendations were shown to the fifth segment which was the control group

Results

The A/B testing showed the following results:

Conclusion

According to the test results, the use of «Alternative and related products» algorithm for BurdaStyle product page increases conversion rate by 27,7%, whereas their average order value slightly decreased by 0,4%. This all together reaches the estimated revenue growth of 27,22%.

BurdaStyle & Retail Rocket comments

«BurdaStyle is beyond just selling clothing patterns online! We strive to provide everything that sewing fans need on our website: patterns, textile, sewing tools, tutorials and inspiration, of course. Professionally crafted Retail Rocket product recommendations is an ultimate solution for achieving this goal: they meet all needs of our clients and help them to find what they are looking for, no matter what purposes customers originally had.

We keep on expanding our service opportunities and brush up customer relations with Retail Rocket. So it is a real pleasure for us to maintain this efficient partnership in the future».

Anastasia Panina, Burdastyle General Manager

«Seeking and exploring new solutions with no fear is a cornerstone of success for this project. BurdaStyle were pioneers in launching mobile product recommendations among Retail Rocket clients. The results we got exceeded all the expectations. It is a pleasure for me to conduct such an interesting and important project».

Anastasia, Retail Rocket Customer Success Manager

 

 

Previous post

myToys.ru online store personalization case study: 10% revenue uplift

Next post

TOM TAILOR twice higher conversion by using Retail Rocket's recommendations

Did you like the article? Subscribe to the newsletter to receive fresh articles in the mail.

Subscribe to the newsletter

We use our own and third-party cookies to obtain statistics of the navigation of our users and improve our services related to your preferences. You can configure your preferences. You can get more information here.