A leading specialty sportswear retailer is testing the viability of artificial intelligence (AI) to aid workflows including markdowns, promotions, and assortments.
A.J. Sutera, executive VP, CIO/CTO of J.D. Sports Fashion/The Finish Line, recently spoke with Chain Store Age about how The Finish Line collaborated with Northwestern University’s Retail Analytics Council AI Lab on a proof-of-concept for using AI and machine learning (ML) in its enterprise operations. The outcomes were so encouraging that the AI Lab spun out a commercial product to AI technology provider RetailPredict.ai.
“About 18 months ago, we started looking at potential use cases, largely around optimizing the more tedious functions of merchandising planning and allocation,” said Sutera. “Looking at the nature of our assortment, the majority is ‘evergreen’ with a predictable markdown cadence. The balance is very volatile and needs more analysis from the planning group.”
As an example of a more “volatile” assortment, Sutera said limited-edition sneaker launches with more nuances present special challenges to Finish Line’s merchandising team.
“There are a variety of colors, celebrity endorsements, reissues of classic models, and limited quantities issued in a narrow market window,” he explained. “You need to determine where to sell and place them, how to price them, and also their availability. Do you only make them available to high-loyalty customers? It’s determined on a case-by-case basis.”
Finish Line also applied AI functionality from the proof-of-concept to optimize aspects of its limited-edition launches including online product recommendation, curation, and arrays.
The retailer worked with Northwestern University’s Retail Analytics Council AI Lab to test the effectiveness of AI technology to optimize markdowns to improve margins by predicting customer demand. Collaborating with Northwestern University Retail Analytics Council AI Lab, The Finish Line tested various algorithms to cluster, classify and predict demand, and then compared the markdown cadence it had historically used for selecting products to a different markdown cadence based on predicted demand, price elasticity and sell-through rates determined with AI analysis. The company found that the model anticipated a potential 8% improvement in gross margin over the historical cadence.
In addition, Finish Line partnered with intelligent personalization platform Reflektion to leverage its real-time Intelligent Personalization Platform to support the layout of product assortments and category pages based on predictive models. Looking ahead, Sutera said Finish Line is focusing on optimizing areas including demand forecast and price optimization.
Currently, the company is also applying AI to determine the best location to fulfill an order, which is not always determined solely by geography.
“We leveraged ML and data science technology to maximize our gross margin while avoiding stockouts,” explained Sutera. “For order management, we are putting smarts on top to avoid markdowns. We want to fulfill orders from the location that will offer the maximal speed and lowest cost of shipment, as well as prevent a product from being marked down. We consider order fulfillment on a SKU-by-SKU basis.”