Fashion Label Uses AI Merchandising To Improve Site Search By 25%

SikSilk, an urban streetwear brand, experienced a 25% increase in ecommerce purchases from on-site search and a 36% decline in site abandonment after implementing AI-powered site search technology.

Nosto’s Head of Search, Antanas Baksys, believes SikSilk is far from alone in failing to use merchandising to optimize search results.

“Our research suggests that around 65% of ecommerce brands don’t currently use merchandising to improve search performance. Merchandising can help you ensure search results are relevant while also highlighting the products you want shoppers to see and act on, such as products with high inventory, trending and seasonal items or related products with higher profit margins.”

The technology from Nosto improved on-site search performance by enabling SikSilk to merchandise the products that are actually in inventory. It not only improved relevancy, but prevented searches in which no products or an incomplete list of products serve up on the ecommerce website.

SikSilk is a British fashion brand with stores across the UK, U.S., Australia and Spain. The company previously relied on the on-site search function of its ecommerce platform. It turned to Nosto to bring more sophistication to the search experience and solve various pain-points.

SilkSilk’s native search technology did not provide the capability to sort products listed in search results based on availability, meaning that products were being pushed that might only have had one available size in stock, for example. This could leave shoppers frustrated, and meant that SikSilk was vulnerable to losing potential sales.

Nosto’s Search now allows SikSilk to create global merchandising rules to ensure all search results boost products with greater inventory and more SKU availability across size, color and style.

Another issue that previously impacted SikSilk’s on-site search performance is that certain search queries did not retrieve all relevant products, restricting the choice of items shown to shoppers and impacting possible conversions. 

Nosto’s search technology uses self-learning algorithms that learn from shopper behavior, so those queries that returned incomplete results -- for example, around color, were quickly solved without needing manual effort from SikSilk.

Some search queries led to no results on the page because shoppers either used a synonym for a product that the previous search technology could not detect or because SikSilk didn’t stock the exact type of product being queried, and didn’t have alternative suggestions in place to show.

SikSilk used Nosto to configure appropriate synonyms for the terms shoppers enter in search queries to fix the issue. It now groups “Jackets” with “Coats” knowing that shoppers used these terms interchangeably. 


Next story loading loading..