An apparel team spotted winners in weeks by ranking products, not forecasting units

New apparel products have only a short window to prove themselves. Many companies rely on experienced sales managers and rule-based reviews, which can arrive too late and with weak accuracy.

The proposed method changes the target. It uses early sales data to rank products regionally and nationally, separating fast, average and slow sellers within two to four weeks of launch.

The researchers validated the model against 2021 orders from a fashion company. When its rankings guided order decisions, the approach produced a 5.9% increase in sales volume.

Predict relative popularity first

Ranking is easier and faster than precise unit forecasting for new seasonal items.

Regional rankings preserve local demand signals that national averages can hide.

Two to four weeks is early enough to change replenishment, pricing and promotion.

The output matches the manager's real question: which products deserve more stock now?

When absolute demand is noisy, a relative ranking can be enough to move inventory toward the right products before the season is lost.

The study positioned ranking-based early identification as a practical decision aid for apparel assortment management. Its validation showed a 5.9% sales-volume lift when the model informed ordering decisions.