Spotlighting seasonal items and the need for forecast value-add analysis

Those who are tasked with forecasting within an organization understand all too well that seasonal items are the biggest supply chain planning pain point. These items are likely to induce service risks or create burdensome inventory spikes due to the high variability in orders. Plus, they make it harder to forecast and properly plan for supply.

While these items do typically have repetitive cycles, there’s increased pressure to accurately forecast cyclical spikes. Here are a few tips on how your organization can improve demand planning efforts for the seasonal items in your portfolio.

Attach a proper baseline statistical model
If you use a seasonal forecasting model for seasonal items, your team can accurately capture year-over-year volume uplifts at the correct time each year. A variety of seasonal models are available to help you create an appropriate forecast. One popular model is the Holt-Winters method. Using this approach, a planner can adjust three different parameters — alpha, beta and gamma — to smooth the base volume, trend and seasonality. Unfortunately, it can be time- and labor-intensive to pick the right parameter values for your model, and you may require additional resources.

Coordinate with your retail partners
A robust sales and operations planning (S&OP) process is important for all SKUs. However, this process is especially key for highly seasonal items because retailers typically only promote these products during the corresponding season. Your baseline statistical forecast may not capture these promotions if they don’t repeat every year. To avoid this costly oversight, work closely with your retail partners to understand what advertisements and promotions they’re running that vary from previous events and plan accordingly.

Measure your forecast value-add impact
After each shipment period, perform a forecast value-add (FVA) analysis to measure your impact. This statistic captures the difference between the accuracy of your baseline forecast and your final consensus forecast. It typically shows significant changes by week or month for seasonal items. While your baseline stat model can capture all your off-month demand, you may still need to add marketing and sales intelligence for in-season forecasts. Tracking your organization’s FVA monthly over time allows you to understand trends and when it’s useful to overlay volumes.

Seasonal items, if not forecast properly, can create significant service risks and waste valuable materials and labor. To layer demand accordingly, establish the proper baseline statistical model for these high variable items. Be sure to include cross-functional insight from within your organization and your retail partners. However, the most critical piece for any organization involves checking your FVA analysis to determine when the baseline model or consensus forecast is more accurate. Doing so is especially important when tracking in-season versus out-of-season demand. Depending on your product portfolio, you may be able to plan sufficiently for out-of-season items using a baseline stat model. If, however, your internal resources are limited, consider using a statistical forecast-as-a-service approach to be sure your time and effort are well spent.