Demand Planning Against COVID-19 – Best Practices

Demand Planning COVID Best Practices Blog - QAD DynaSys

This is the last of this series of three articles related to the impact of COVID – 19 on demand planning. After an interesting introduction, the second article talked about the impact of COVID-19 for demand planners. This third article focuses on proposing a set of best practices that planners can carry out to improve forecast accuracy in moments of great uncertainty.

The first thing we must do regarding demand planning in this unprecedented situation is to adapt our processes and systems to track and anticipate abrupt changes in demand much more quickly than in times of stability. For statistical forecasting, we should consider using models and parameters that are more reactive to recent sales history. This will mean more interaction with the models and more overrides as the demand planners adapt the Forecast to new scenarios.

Taking into account what was previously argued, the first good practice would be to increase the frequency of the forecast review cycle with the focus on a short horizon. Instead of working on monthly horizons, we should reduce to weekly horizons in order to monitor and adapt more quickly to gain agility and responsiveness in scenarios where demand experiences significant deviations from the forecast. Although the forecast may experience large deviations, it is not recommended to deactivate the statistical calculation as it may require a lot of user interaction. The statistical forecast will still provide insight, and eventually new trends. It also provides a basis from which to measure variations. As it adapts to the situation, it will still serve as an initial proposal, but manual adjustments will be needed to increase accuracy during the disrupted period.

Regarding forecast corrections, it is highly recommended to work with adjustments at aggregate levels. QAD DynaSys sees that forecast errors are not currently the result of specific situations, but of wider external factors. These factors impact a large number of forecasts by customer, channel, country, product type … For this, software providers must collaborate to offer simple solutions that are tailored to customer needs. For example, we might need to turn off a customer’s impact on Forecast because of temporary business shutdown or even failure. We may have to reduce a specific customer’s outlook because physical shops are closed and the only open channel is eCommerce. Capturing the change in the channel, in this case, is important too, as it is likely to have an impact on distribution routing

In any situation in which uncertainty is very high, we must focus our resources on items that have the most impact on the company and where problems or deviations could be costly in economic and strategic terms. ABC classification is highly recommended to help the business concentrate on the right areas. However the ABC is defined (by volume, revenue, margin, strategic importance…), now may be a good time to review the ABC to ensure it is still current and relevant. This will help our demand planners focus on the correct “A” products. At a time when Demand Planners are stretched, it may be helpful to bring other people in to help. If they are less experienced, they can be allocated B and C products.  

Another recommendation is to reduce the forecast horizon. Currently, companies might work with a horizon longer than 18 months. QAD DynaSys recommends modifying the horizon to a shorter one and focusing on three different periods. The first, and main focus, would be short term, for instance, a weekly horizon of 0 – 12 weeks to cover the first part of the ongoing COVID-19 phase which will probably be the most uncertain phase with the biggest fluctuations. When reviewing this phase, external factors and leading indicators may need to be considered more than is normal. Secondly, define a term between 3 – 6 months as the medium-term where we can maintain the S&OP process taking into account that the most important factor is to keep the process as agile and adaptive as possible. Finally, especially for companies with marked seasonality, it is also recommended to work at a horizon of 7 – 12 months to see if the seasonality of these products has been affected by the emergence of COVID-19. The disruptions we are currently experiencing will need to be addressed in terms of their impact on seasonality in the future.

A fundamental practice is the use of alerts and forecast monitoring tools (KPIs). Businesses have to monitor the forecast and compare it with actual demand identifying where the largest deviations are and react accordingly. Continuous monitoring will allow companies to be ready for changes, detecting them quickly, thus adapting processes to new scenarios. Using error indicators such as MAPE or Forecast accuracy will help us to easily understand the issues facing our business. Bear in mind that forecasts will always have errors so we have to define a tolerance to errors that is right for the current situation. We may need to look at our KPIs and decide whether we are using the right ones for managing the current situation. In most cases, this will mean putting more focus on the short term. A well-balanced set of KPIs should contain a relevant mix of long and short term measures already

To sum up, it is important to be aware of the exceptional position COVID-19 has put us in and try to use all resources, both internal and external, to be able to face this situation. This may mean putting more people into the demand process. Starting from the premise that the forecast is wrong will push companies to be proactive with the results, questioning them and taking the necessary actions so businesses can benefit from the forecast and get the greatest value possible with the highest accuracy.

Taking into account all the points presented in this article, QAD DynaSys is committed to offering its customers the necessary tools so they can benefit from the results and agility of the software not only during the pandemic but also for the subsequent analysis.