A real-world case study using SAP Analytics Cloud to forecast product sales across stores and countries — minimising MAPE through rolling-window analysis and observation tuning.
In today's competitive landscape, anticipating future trends is essential for strategic planning. With SAP Analytics Cloud (SAC), businesses can leverage powerful time-series forecasting to gain actionable insight. This piece presents a practical implementation predicting sales across product families through SAC's predictive scenario capabilities.
The business scenario
The project aimed to forecast sales for each product across two store chains in different countries, using a dataset spanning January 2017 to December 2020 with store, product and daily sales records — predicting sales for the next 15 days for each store-product combination. Accurate forecasting is critical for optimising inventory, planning operations to regional demand, and supporting budget allocation and promotional effectiveness. The team minimised Mean Absolute Percentage Error (MAPE) — the percentage difference between actual and predicted values.
Step-by-step in SAC
Historical data was structured by date, product, store and sales, then cleaned (date handling, outlier removal). SAC's predictive scenarios were applied with country-specific analysis and observation-based insight — testing weekly, monthly and aggregated windows. Rolling-window analysis captured short-term trends while maintaining broader patterns; the lowest MAPE came from a 15-day window with a 6-month size. Optimal outputs then fed SAC Planning Scenarios for what-if analysis.
Result analysis: component impact
Trend accounted for 87.28% of variance — the dominant factor reflecting long-term direction. Cycles contributed 6.13%, including a weekly cycle of periodic fluctuation. Final residuals were just 6.59%, demonstrating model robustness.
Weekly cycle breakdown
Saturday showed the highest positive impact, followed by Sunday — aligning with weekend shopping. Monday and Tuesday showed negative impacts (post-weekend dips), with Friday acting as the transition into the high-impact weekend. The implications: optimise inventory and staffing for weekend peaks, use midweek promotions to counter Monday–Tuesday dips, and leverage Friday's pre-weekend momentum.
Conclusion
SAP SAC is not just a forecasting tool; it's an integrated platform connecting predictive analytics, business planning and intelligence in one solution. Working with historical data, applying advanced models, and visualising results empowers genuinely data-driven decisions.
