Predicting Sales Cycles

Weekly revenue directly reflects business performance and responds strongly to seasonal phases. Tracking this metric provides the clearest view of sales patterns, enabling better planning for promotions, inventory, and staffing compared to relying on total orders or average order value alone.

Sales follow clear and repeating patterns rather than random fluctuations. Seasonal cycles consistently lift revenue during peak weeks and reduce it during slow periods. The small residual effect confirms that most variation is explained by seasonality and trend, not randomness.

Revenue naturally falls into Peak (32 weeks), Moderate (41 weeks), and Slow (32 weeks) phases. This classification enables targeted strategies: concentrate promotions and stock-ups in Peak weeks, maintain steady operations in Moderate weeks, and implement cost-saving campaigns in Slow weeks.

Seasonal timing is the strongest factor affecting revenue. Peak Season weeks strongly correlate with higher revenue, while Slow Season weeks link to lower revenue. The passage of time itself has minimal effect, proving that seasonal preparation is the most effective revenue management lever.

The model was tested on 21 weeks of unseen data (20% holdout), successfully capturing peak and slow patterns beyond the training period. This validation confirms that seasonal cycles are not only historical but also predictive of future weeks, providing confidence for forward planning.

Regression analysis provides concrete numbers: Peak Season weeks increase revenue by approximately RM8,430, while Slow Season weeks reduce it by around RM7,914 compared to Moderate weeks. These statistically significant values enable realistic target setting and budget planning.

The model explains 78% of weekly revenue variation, with both Peak and Slow Season effects statistically significant. While not precise for exact week-to-week predictions, it provides a strong guide for understanding when sales will rise or fall, supporting strategic decision-making.