Inventory Management and Demand Forecasting

Textile inventory management balances stock availability against working capital cost and markdown risk across a product range characterised by short selling seasons (8–12 weeks for fashion items), high demand uncertainty (sell-through CV% 40–60%), and long procurement lead times (8–18 weeks from Asia) — creating structural tension between fill rate targets (95% in-stock) and inventory efficiency (sell-through >75% at full price). Open-to-buy (OTB) planning: financial budgeting system allocating purchasing budget by category, month, and vendor — OTB = planned sales + planned ending inventory − beginning inventory − on-order; OTB discipline prevents over-buying that leads to excessive markdowns (US fashion industry average markdown rate 32% of original retail price, representing $500 billion annual value destruction). Demand forecasting methods in textiles: statistical forecasting (moving average, exponential smoothing, ARIMA — 55–65% forecast accuracy for fashion, 75–85% for basics); machine learning (neural networks, gradient boosting using POS data, social media trend signals, weather data — 70–80% accuracy improvement over statistical methods for AI adopters per McKinsey 2023); consensus forecasting (commercial team + buying team + historical analogues — most common, accuracy 60–70%). Safety stock calculation: SS = Z × σ_D × √LT where Z = service level factor (1.65 for 95% in-stock), σ_D = daily demand standard deviation, LT = lead time days — cotton T-shirt basics (LT 90 days, σ_D 50 units) requires SS = 1.65 × 50 × √90 = 782 units safety stock per SKU. Fast fashion demand response: Zara's biweekly store delivery replenishment system uses 2-week rolling POS data to adjust production of committed designs — committing only 50–60% of seasonal volume upfront versus industry average 80–90%, preserving 40–50% budget for in-season responsiveness that improves sell-through 12–18 percentage points versus traditional planning.