Sales Forecasting

The business question

How much will we sell next week, next month, next festival season? The answer drives purchasing, stock levels, staffing and cash flow. Over-forecast and money sits in stock (or, for perishable goods, goes to waste); under-forecast and sales are lost.

What I built

• Cleaning and resampling of historical sales data, including gaps and outliers
• Trend and seasonality analysis: weekly patterns, months and festival peaks
• Forecasting models evaluated against a simple baseline ("same as last period"), because a model that cannot beat the baseline should not be used
• Visual forecasts that a manager can read without a data background

How it is used in a business

• Purchasing and stock: order quantities based on expected demand rather than habit
• Cash flow: expected revenue by week, for planning payments
• Accountability: forecast error tracked every period, so everyone knows how far to trust the numbers

How success is measured

Forecast error (MAPE) compared with the baseline, and in operation, stock-outs and leftover stock before and after the forecast is adopted.