Apparel : Merchandise Intelligence

The apparel industry is driven by qualitative elements such as sense, aesthetics, intuition, branding, and trends, which are difficult to quantify or explain clearly. With new products launched twice a year for the S/S and F/W seasons, the pace of change is unparalleled. The focus varies slightly depending on business models such as manufacturing-distribution, direct-buying, or consignment sales, which also leads to differences in KPIs. Companies that combine manufacturing and distribution, for example, target a sell-through rate of around 66-75%, considering sales at outlet stores. In contrast, pure-play retailers specializing in direct-buying or consignment sales aim for a 100% sell-through.

The role of a merchandiser (MD) is to lead the entire process from planning seasonal items to final sales. Typically divided into S/S and F/W seasons, the MD plans the season's items on a six-month cycle and begins sales as the season starts. Based on the company's sales targets, MDs develop strategies for their assigned items, conduct market research, plan products, and discuss marketing and promotional strategies. The initial inventory for the first 2-2.5 months of the six-month period is ordered, determining how much to produce/purchase in different colors and sizes. This is known as assortment planning. Since the apparel industry typically has a lead time of around 2-2.5 months, MDs monitor sales trends, inventory levels, and lead times to reorder the right quantities at the right time.

Understanding this process helps to explain why applying AI to the apparel industry has been challenging. Traditional AI-driven demand forecasting relies on predicting sales volumes based on historical sales data. However, as shown, the MD's decision-making process is much more complex, intuitive, and sometimes even determined. While MDs do refer to past sales data of similar products to develop strategies, the apparel industry heavily depends on the experience, sense, and intuition of experts in the field.

DEIN Station's Merchandise Intelligence focuses on supporting the decision-making process aligned with these MD tasks. Unlike AI, which claims precise predictions, Decision Intelligence emphasizes collaborative decision-making between AI and MDs. For example, decision models developed using data on attributes and sales history for each SKU provide recommendations on production quantities, optimized assortments, and reorder timing/quantities. MDs can then refer to this information to make more refined decisions. The recommendations from decision-making models, the MD's decisions, and the deviations in actual sales are continuously captured by the Merchandise Intelligence model, leading to progressively refined intelligence over time.