How the Kant sporting goods store configures useful product recommendations

25 Sep ‘19

KantKant is a sporting goods store for outdoor enthusiasts. They have 600 thousand customers shopping online, and 20 stores in 6 cities across Russia. The site is based on Bitrix, while the back office is 1C.

Kant flexibly adjusts the algorithms of related, similar and popular products, providing customers with the most relevant recommendations. Let’s share some details behind those settings. The opinion on the results presented here was prepared by Anna Belovodskaya, Director for Electronic Commerce and CRM, together with a Kant marketer.

Situation

Before switching to Mindbox product recommendations , another service was used to present on site recommendations. Which left much to be desired! For instance, the issuance of product recommendations could not be tuned by hand, so, that led to not the most useful stuff popping up as recommendations to be purchased alongside sporting equipment. For example, we wanted to recommend a protective helmet to customers who were focused on snowboards.

We switched to Mindbox recommendations because of the provided ability to manage product recommendations through their interface and suggest the best stuff to customers.

Anna Belovodskaya Anna Belovodskaya,
Director of Ecommerce and CRM at KANT

Kant opinion

«In many categories of mass market products, it is possible to work with product recommendations through the use of big data technologies , via the capabilities provided by predictive analytics. But if we’re talking about professional athletes and sports enthusiasts of an advanced level, then the technical characteristics of the goods, compatibility, and technology get shifted to the front of the line. And here, recommendations from the «black box», which represent the majority of recommendation solutions existing in the market, are inferior in quality to the recommendations set up by taking expert knowledge relative to an item into account.

The Mindbox platform allowed us to set up cross-category recommendations in the most optimal way for our segment».

How we launched on site product recommendations

In total, three Mindbox product recommendation algorithms were installed on the Kant website:

  • Related products are displayed in product cards.
  • Similar products display in product cards and search results.
  • Bestsellers are placed on the main page.

We’ll tell you more about each of these algorithms: what they look like, recommend, and how they’re configured.

Related products are displayed in product cards.

When an athlete purchases equipment, such as ski’s, well, some ski boots, a protective helmet and warm socks will definitely come in handy. The automated algorithm, however, will not be able to take into account all of the intricacies related to sports equipment.

Related products from appropriate categories are selected by manufacturer, season, gender, age, and price range. Detailing at this level allows one to select product recommendations of the highest quality.

To compliment hiking boots, we would recommend tents, hiking backpacks, Swiss knives, a headlamp. For running sneakers – a running jacket, sunglasses, a heart rate transmitter.

Related products

 The Mindbox interface is set to manually match one category to another, which allows one to deliver more accurate recommendations, while retaining all product specifications in mind.

The Mindbox interface

Similar products displayed in the product card

We recommend other products which are similar to ones that a client is viewing directly in product cards. The algorithm takes into account collections, seasons, model types, categories, price, gender, age, and manufacturer.

Similar products

The similar product algorithm takes into account 9 independent properties of a product. A products position in the column indicates the priority of a parameter in the definition of similar products.

Algorithm settings in Mindbox are as follows:

Algorithm settings for similar products for the site
Algorithm settings for similar products for the site

Bestsellers on the main page

On the main page, we recommend the most popular products based on on-site sales. Kant utilizes Mindbox as a CDP, so we receive information pertaining to all sales made through the website.

The bestsellers

Based on the online store’s complete sales history, the popular products algorithm, based on sales, makes a selection of the best-selling products and displays them on the main page. Within the settings interface of the bestseller algorithm, one can even filter what products are to be displayed:

The settings interface of the bestseller algorithm

Conclusion and what’s to be expected next

We set up and launched three on site product recommendation algorithms. Currently, we are testing the similar product algorithm in the product card relative to the control group. This is done in order to reliably measure the added revenue provided by this algorithm.

Kant Marketer opinion

«High expectations were in place for Mindbox product recommendation functionalities .These expectations were met! In our segment, compiling sports equipment product selections based on statistics only is not enough.

The following aspects are critical for consideration: expert recommendations, the level of training and, as a consequence, the price range of equipment. Special thanks to the Mindbox team needs to be given for their attention to detail and the desire to take into account and implement our wishes relative to recommendation functionality.

We plan on improving our quality and developing this direction in the future».

«KANT» Marketer

Материал подготовили

Elizaveta Kurochkina

Elizaveta KurochkinaProject Manager

Igor Kalinovsky

Igor KalinovskyProduct recommendation product owner

Philip Volnov

Philip VolnovAuthor

Subscribe to our quarterly news digest

Tell us a little about yourself

We’ll respond within 24 hours

Partnership request

Typically we’re answering within 24 hours