Forecasting the Easing of COVID-19 Related Measures has been saved
Forecasting the Easing of COVID-19 Related Measures
“History only tends to repeat itself when history is stable”
We are currently living in uncertain times and can only speculate on when certain measures will be relaxed or abandoned. As these measures impact most of our businesses we need to find a way to deal with this and make sure we can make the right business decisions, now and in the future. This article outlines an approach that helps forecasting during these unpredictable times.
The increased importance of drivers
There are many different methods for drafting a forecast. Many companies take last year's figures and make judgment calls on the adjustments necessary. Other companies adopt a non-statistical driver-based approach and where necessary add judgment calls. More advanced forecasting is done by using regression analysis based on statistical inference. All of these methods are affected by the COVID-19 crisis and need to be revised as a result of that.
Adjusting incremental budgets for the COVID-19 impact with the easing of measures insight is very difficult. Incorporating drivers into an organization’s forecast can support making adjustments as new developments take place. Be sure to test the extent to which these drivers predict the outcome. In normal circumstances, one could use a regression model for this, but given current times expert judgment is increasingly important because history only tends to repeat itself when history is stable.
Selecting forecasting drivers
Starting with observing how well drivers have historically predicted your revenues and costs will give an idea of how well they work. Good data granularity, such as sales calls, call length, web traffic, conversion rates, and logged opportunities, are good starting points to explore. Also, try to isolate recurring costs and revenue from variable/incidental revenues by using different drivers for these. Lastly, try to brainstorm on what external data might help predict your sales and costs and see whether that helps to explain part of the variance.
Plan for various scenarios
Including different perspectives makes for a better forecast. Especially during these uncertain times, it is wise to have clarity on how various scenarios, from worst to best case, would impact your business. By doing so, management can make more balanced decisions compared to decisions made based on a single forecast that averages all these scenarios. For example, a company facing a major investment opportunity might conclude from a single forecast that it is a profitable decision to invest. However, after breaking down the analysis into different scenarios, it turns out that the worst-case scenario, where the company will face severe financial issues, is probable as well. When using a scenario-based approach, make sure to clearly outline assumptions on which each scenario is based.
Be sure to keep score
The Plan-Do-Check-Act cycle is probably known by all financials and most other disciplines. However, within forecasting analysis on deviation from the forecast is often performed superficially. This is unfortunate because there is a lot to learn from observing how forecasts play out in the real world. Keeping score can help to recalibrate your forecast by changing the strength of relationships and eliminating drivers that have little predictive power. COVID-19 has a serious impact on all of this. Even if you used to have a well-calibrated model, keep testing the effects and relationships. Where necessary, adopt a different forecasting model from what you are used to, accommodating this new reality. In all circumstances: Challenge the model. What has worked well historically is not guaranteed to work well today, so keeping score is crucial.
Erik Gielstra, Deloitte Business Process Solutions (2020)
Want to know more about our view on forecasting the easing of COVID-19 related measures? Explore business opportunities together with Esther Roelofs (+316 217 00 347 or email@example.com) and Martijn Kösters (+316 227 54 158 or firstname.lastname@example.org).