Forecasting for uncertainty
With the global pandemic sending shockwaves through markets, most companies are facing significant changes in their demand patterns. These sudden and unexpected drops and peaks in demand require a rapid response here and now. But they could also potentially haunt you in the future, as this black swan event disrupts your historical sales data and the foundation of your statistical forecasting.
This article will provide you with insights that can help you track your forecast right now in the midst of uncertainty, in a time when you can’t lean on historical data. You’ll also get an idea of how to look towards future forecasts and work around the big pattern shifts created by the COVID-19 pandemic.
Amidst new regulations and ways of interacting in society, one thing is clear: consumers are spending more time at home. In other words, consumer demand is changing vastly, wreaking havoc on supply chains in the business world. As projects and orders are postponed and cancelled, it’s crucial that companies get ahead of this ripple effect. This means that key account managers need to get even closer to their customers and truly understand their mindset and needs.
Updates on changes in demand need to occur more often, and your usual monthly S&OP cycle will need to run more frequently. There has never been a more opportune time to focus on also running an S&OE cycle to ensure that Planning, Operations and Sales are all aligned and have the most up-to-date information.
No one can predict the future. So, you need to be prepared to make plans alongside the many uncertainties and unanswered questions. Scenario planning can help you mitigate the risk attached with uncertainty, allowing you to test how your organisation will react to different factors in your forecast.
Scenario planning is tool that helps you make decisions to counter and/or exploit upcoming events, such as the COVID-19 situation, based on qualified assumptions and identifying drivers and possible outcomes.
You can read more about IBP/S&OP scenario planning and our five-step framework for using it here.
It’s difficult to see a general pattern in how companies are impacted by the global pandemic. Some product categories have been reduced to almost nothing, while other categories are enjoying a high increase in demand. For example, while FMCG players experience little to no demand from restaurant and catering companies, the demand from supermarkets is markedly higher.
With varying supply chain setups between product categories and business units, it’s important to understand the impact of changes in demand and its extent. This makes it easier to identify a focus point for mitigation activities. Measuring the short-term bias across your portfolio is a simple way of gaining an overview, allowing you to quickly see which areas are impacted, while also providing you with an understanding of quick actions that can be taken, e.g. stopping production, reducing raw material supply, scaling down blue collar workers etc.
Forecasting is extremely difficult during this period of rapid change. One way of tackling this challenge is adapting the rest of the supply chain to the current situation. Structure your planning by respecting the unpredictability in the market. For product segments that are impossible to forecast for, consider consumption-based planning (Reorder Point (ROP), DDMRP) or an ATO/MTO strategy. Although this is not a new method, fast-changing demand may require ATO/MTO on a large range of your portfolio.
Engaging in a dialogue with your customers about a temporary increase in lead time tolerance could potentially also allow you to start producing some of your products as “assembly-to-order” instead of “make-to-stock”, meaning less reliance on forecasting. This in turn minimises the risk of scrap and frees up networking capital by reducing your stock levels.
Instability and uncertainty are unpredictable, and it’s impossible say whether the world will return to the markets of 2019 or evolve to something entirely new. One thing we do know is that the historical data we usually base our forecasting on will not fit the regular pattern, when you begin running statistical forecasts in the future.
These data anomalies will require rigorous data cleaning and a thorough re-evaluation of seasonality trends. Here, you should consider two factors:
There are many methods and techniques for erasing data within a certain timeframe, ranging from simply ignoring the history, adjusting the mean based on pre-pandemic periods to reusing history from 2019.
If the changes are permanent, the historical data will ultimately become less relevant for your future statistical forecasts.
“It is very hard to predict, especially the future.” The words of the Danish physicist Niels Bohr are relevant now more than ever when it comes to forecasting. In times of uncertainty, keeping your customers close and maintaining frequent contact points with the rest of your supply chain is crucial to staying ahead of demand. A full transparency approach will also help you gain an overview of changes and a solid understanding of how different product groups and business units are affected.
However, the key to tackling the rapidly changing supply and demand landscape lies in a future-oriented mindset. Everything is temporary – including this pandemic. It’s important to keep in mind that once this is over, we’ll have a new challenge to tackle – forecasting and planning with potentially corrupt historic data. We may not be able to predict what’s coming next, but we can use the tools and methods at hand to prepare and plan for the uncertainty ahead.
Part 4: How did the Implementor turn out? What we have learnt, and what it might be used for some day.
Implement Consulting Group