Vector (43)CASE STUDY — ADVANCED CATEGORY NAVIGATION

Leading sports retailer in Europe increase revenue with 12% with Loop54 Advanced Category Listings

DOWNLOAD CASE STUDY PDF

SPORT

“In the past, we have always displayed our campaign products on the top of the category listings page. That’s not optimal for all individual customers, so we wanted to see if we could offer the user a more personalized and optimized presentation of the items within each category. We are very pleased with the results that the “Order & Structure” setup of Loop54's Advanced Category Listings gave us. The fact that we now can personalize our product listings and also provide an optimized structure on the site will give customers a better shopping experience online.

Kim Andre Nilsen
VP eCommerce & customer experience at XXL

About XXL

XXL’s vision is to be the preferred retailer for sports and outdoor activities in Europe. Since its launch in 2001, XXL has turned into a company with an annual turnover of approximately NOK 10 billion (≈ EUR 1 billion). Today XXL has 90 stores in Norway, Sweden, Finland, and Austria and is also the largest online sports retailer in the Nordics.

 

In total, XXL has more than 5 000 enthusiastic employees and offers its customers a unique mix of great brands, great expertise, great assortment, great accessibility and great prices.

 

Background to Category Navigation project

At Loop54, we continuously work with our customers to further improve the performance of their search and category navigation. XXL had successfully been using Loop54’s eCommerce search and personalization since 2015, when we suggested testing Loop54’s recently launched Advanced Product Listing Pages against their current solution for category listings based on Apache Solr. Our ambition was to extend the relevant results we were providing for searching visitors to visitors that are browsing categories to find the right products, while increasing conversions and revenue for XXL.

 

Advanced Category Listings

Based on the latest version of our AI algorithm, Gas GOLEM, our Advanced Category Listings benefit from an even better understanding of the relationships between products in your catalog, as well as which products best fit each visitor’s current intentions. Instead of manual merchandising, our algorithm automatically ranks and sorts category listings based on factors like general popularity, personal relevance, and business logic. Our product specialists fine-tune the combination of these to find the best engine setup for each individual store.

 

 

Pic (1)

The hypothesis of how people want to navigate categories

How visitors navigate categories in an online store varies greatly, not only between different industries but also within, depending on factors like UI design, the size of the device the products are displayed on, and individual user preferences, amongst others. To provide the best setup for each of our customers, or product team developed different hypotheses of user behavior and corresponding engine setups.

 

For XXL, we configured three different variants of the engine to compare to their existing solution for category listings.

A - Hypothesis - Discovery

The first hypothesis is based on the assumption that the user navigates a category listings page by using facets to narrow down the scope of displayed products. Instead of grouping the products into clusters of the same type, we wanted to maximize the exposure of different product types high up in the listing to give the visitor a sense of the variety of products that the category contains. A category listing for women’s clothing would, in this scenario, have displayed a mix of popular trousers, tops, jackets, and boots at the top of the results. The most prominently featured products from each type cluster were determined by a mix of general popularity, personalization, and boost & bury rules.

The results

I
+

0

%

Average Order Value

I (1)
+

0

%

Average Session Value

I (2)
+

0

%

Average Price

I (3)
+

0

%

Revenue Increase

Pic (2)

B - Hypothesis - Lightweight Lucy

For our second scenario, we used the same business logic as XXL’s existing solution, testing the hypothesis that boosting campaign products will yield the best results for the retailer. Therefore, we created a stripped-down setup for category listings that ranked products according to a combination of general popularity in the category and overall popularity of the products while boosting campaign products.

The results

I
+

0

%

Average Order Value

I (1)
+

0

%

Average Session Value

I (2)
+

0

%

Average Price

I (3)
+

0

%

Revenue Increase

C - Hypothesis - Order and structure - Winner

The “Order & Structure” hypothesis assumed that visitors scroll through category listings from top to bottom to find the products they are interested in. For this setup, we grouped the different product types in each category into contextual clusters and sorted the various groups from most relevant to least relevant context, based on the intent the user had shown. In this scenario, a category page for women’s clothing might have listed all pants first, followed by all jackets, shoes, etc. Within each product group, the products were sorted based on a combination of popularity, personal relevance, and boost & bury rules.

 

Developing this setup further, products in a cluster can also be further divided into sub-groups using, for example, price or other ways of differentiating the products.

The results

I
+

0

%

Average Order Value

I (1)
+

0

%

Average Session Value

I (2)
+

0

%

Average Price

I (3)
+

0

%

Revenue Increase

Pic (3)

D - Boost of campaign products (XXLs existing solution based on Apache Solr)

As a reference setup, we used XXL’s existing solution that applied business logic in the form of boosting campaign products.

 

Conclusion

For XXL, the Order & Structure setup performed the best on all metrics during the test period and resulted in a revenue increase from category listings of 12.68 % compared to XXLs existing solution. The overall revenue increase was 4,5%. Lightweight Lucy also performed well on revenue but with a decrease in AOV.

 

After completing the A/B/C/D test, XXL decided to continue with the Order & Structure setup for their category listings.

Ill_07 1

Loop54 offers true personalised on-site product search and category navigation

Shoppers expect the same level of relevance and personalisation online as they experience in-store. Powered by Machine Learning and built exclusively for eCommerce, Loop54 delivers that exceptional online shopping experience.

  • Automated: Automatically learns words and merchandises search and category listings
  • Relevant: Interprets search intent to deliver truly relevant results
  • Personalised: Sorts results according to popularity and personal taste

REQUEST A DEMO TODAY!