Forecasting sales for both fast- and slow-moving items presents unique challenges due to the differences in how demand fluctuates. Fast-moving items tend to have more predictable sales patterns, allowing for precise forecasts, while slow-moving items often experience higher relative forecast errors, even when absolute errors are similar.
Key Points to Consider
1. Relative vs. Absolute Errors:
Forecasting for slow-moving items results in higher relative errors compared to fast movers. For example, predicting the sale of a single slow-moving item daily is inherently less precise than forecasting hundreds of fast-selling items.
2. Avoid Naive Scaling:
Simply scaling down forecasting methods for fast-sellers to accommodate slow-moving items doesn’t work. The error behavior differs across scales, and slow sellers have much more variability in demand, which needs to be considered.
3. Scale Awareness in Forecasting:
When planning stock, make sure aggregation times align with business needs. For example, daily forecasts for non-perishable, slow-moving items may not be suitable. Longer-term aggregation, like weekly forecasts, often yields more reliable results.
4. Practical Implications:
Forecasting for slow-moving items improves when aggregation occurs over longer time scales or for higher-selling items. Daily fluctuations cancel out over time, leading to more accurate forecasts. This is particularly useful for reducing waste in perishable goods and avoiding stockouts in other categories.
5. Forecasting Accuracy and the Poisson Distribution:
For items with low, unpredictable demand—like sea cucumbers—forecast accuracy is limited due to the “Poisson distribution” behavior. Small-scale forecasts are inherently noisy, with some days having no sales and others seeing unexpected demand spikes. This makes it unsustainable to stock such perishable items.
6. Evaluating Forecast Accuracy:
Rather than applying a blanket precision target, forecasts should be grouped by predicted sales rate. This stratification accounts for natural demand variability, especially for low-demand items, ensuring forecasts more accurately align with actual sales patterns.
Key Takeaways –
Match Forecast to Decision-Making Timescales: Use daily forecasts for perishable items but shift to longer-term aggregations (weekly, monthly) for durable goods or slow movers.
Set Rate-Dependent Targets: Establish forecast accuracy expectations based on the predicted sales rate rather than using a single threshold for all products.
Account for Noise in Slow Sellers: Acknowledge that slow sellers inherently carry more significant forecast uncertainty and consider strategies like safety stock or produce-upon-order to mitigate risk.
By understanding these forecasting principles and adjusting for the differences between fast- and slow-moving items, companies can better align stock levels with actual demand, reducing waste for perishable items and ensuring that stockouts are avoided for all products. Strategic adjustments to forecast evaluation, aligned with the expected sales rate, can significantly improve business decision-making.
