Predicting Demand to Power Decisions
Objective
To forecast product demand and optimise inventory levels across stores and regions while accounting for seasonality and promotions.
Business Problem
The retailer faced costly overstock and stockouts due to reactive inventory planning. Promotional campaigns often disrupted baseline demand, leading to waste and lost sales.
Analytical Approach
- Built regression and time-series models (SARIMA + Prophet) for 120 SKUs across regions.
- Integrated weather and discount variables to model demand elasticity.
- Compared forecast accuracy before and after model implementation.
Key Metrics
- Forecast Accuracy Improvement: +17%
- Stockout Reduction: −12%
- Promotion ROI Increase: +6%
- Sales Forecast: $620K CAD per quarter
Insight
Demand was highly seasonal with strong autumn peaks. Certain promotions cannibalised future sales rather than generating incremental revenue. Optimising discount timing produced significant margin gains.
Business Value
Demand forecasting empowered supply chain teams to align inventory with reality — freeing working capital, increasing sales reliability, and enhancing customer satisfaction.
💡 Coming Soon: Synthetic dataset, forecasting notebook, and Power BI dashboard.
Tags: #RetailAnalytics #Forecasting #MensahInsights