French sporting goods retailer Decathlon launches a U.S. e-commerce site and uses a machine learning-based site search tool.

Sporting goods retailer wanted to give its new U.S. e-commerce site a winning start.

The retailer launched its first e-commerce site for U.S. shoppers in December 2017 in conjunction with opening its first U.S.-based store. French-based Decathlon has sold sporting goods since 1976. The retailer has 1,200 stores in 50 countries and operates roughly one e-commerce site for each country.

With this being Decathlon’s foray into the U.S. market, which has the potential to be large and lucrative, the retailer wanted to be sure it nailed its e-commerce site, says Tony Leon, Decathlon’s chief technology officer.

The retailer spent a year creating its U.S.-facing site and realized that its on-site search was not as robust as it could be, Leon says. Even though the retailer has several e-commerce sites and site search already built for its 40,000 SKUs, how shoppers search for products varies by country, he says.

For example, in the U.K., consumers may search for “trousers” and “football,” whereas U.S. shoppers are looking for “pants” and “soccer.” Besides cultural differences, consumers search for products differently, Leon says. Three shoppers may all want hiking shoes, but one shopper may type in the search bar, “I want to go hiking,” another “shoes for mountains” and another “hiking gear.” Decathlon wanted its site search to work well for all types of queries.


“We all have different ways of expressing what we are looking for,” Leon says. While the retailer could re-tag all of its products, that would be a time consuming, manual process that could still miss what the shopper seeking, Leon says.

The retailer decided to use a machine-learning algorithm for its site search with artificial intelligence search vendor Adeptmind. The technology enables the retailer to feed the search engine products and the engine learns based on what shoppers look for, click on and purchase. Plus, it learns without being programmed. For example, instead of having to tag a T-shirt that is “red” with every variation of red such as cherry, crimson or maroon, the algorithm will be able to learn that all of those colors are similar and associate them with a red search query.

The time it takes from when a shopper lands on to making a purchase decreased by 48% between January and April. Additionally, a shopper’s exit rate after using site search has decreased 63% during the same period. In addition, its search conversion rate, or conversion rate from when a shopper uses the site search bar, has increased more than 175% in this time period, Decathlon says. While has had the site search tool since it launched its U.S. site, this shows the tool continually learns and improves over time, the retailer says.

The technology also saves the retailer on resources. In Europe, Decathlon has employees dedicated to analyzing search queries and updating its site search, and Decathlon doesn’t need these dedicated employees in the U.S., Leon says.

“It’s less work day to day and for better results,” Leon says about the tool.


In these first six months of operation, about 20% of Decathlon’s U.S. sales are from its e-commerce site, and the remaining 80% are via its one store. This 20% is roughly four-times greater than its other e-commerce sites. Leon attributes this to the mature online shopping market in the U.S.

Even though U.S. shoppers are comfortable with buying products online, Decathlon still wanted to open a physical store. Leon says it’s important to meet consumers physically, and some shoppers like to come in, ask questions and touch the products. Plus, having a physical presence keeps its brand visible.

It took between two and three months to implement the site search feature on the e-commerce site, says Adeptmind’s co-founder G Wu.

Adeptmind collects “a small percentage” of each retailer’s monthly sales that stem from site search as its fee, and for small retailers, this is typically a few hundred dollars per month, Wu says.

Adeptmind also is working to create a plugin for merchants that use Shopify Inc. (No. 8 in the 2018 Leading Vendors to the Top 1000 for the e-commerce platform category), BigCommerce (No. 7) or Magento (No. 2) e-commerce platforms.