6 forecasting best practices from BetterVu

Now that we are entering the second quarter of the year, many finance teams are getting down to their first systematic reforecast, while more agile organizations that adopt a monthly rolling forecast may have already completed two or more. In our fast-moving business world, generating more frequent forecasts can bring obvious benefits. But according to recent research from Deloitte, only about a quarter of businesses practice rolling forecasts—and those that do tend to be smaller businesses. So what’s holding back the majority?

Towards the end of last year, we asked Mitch Max of Anaplan’s Toronto-based global partner BetterVu to share some best practices on how to produce more timely and accurate forecasts—without the heavy lifting. This last point—avoiding heavy lifting—is critical, Mitch said, because it’s one thing to build an effective forecasting process, but that process is unlikely to be sustainable unless it’s also efficient.

Here, in summary, is what Mitch recommended:

• Identify business drivers that are both predictive and causal, then forecast them at appropriate levels of granularity and dimensionality. For instance, this might mean forecasting retail revenue by sales by square foot at department and store level, and then applying statistical techniques to account for seasonality.
• Make forecasting a continuous process by building driver-based models that can be quickly and easily refreshed with transactional data. That way, new forecasts can be generated in just a couple of days. Mitch stressed that the forecasting process has to be fast if it is to help managers to reach quick and timely decisions that will optimize financial performance in the short term.
• Automate the process while allowing manual overrides, as that’s the only way you can manage the large amount of data involved in such a short time window. This may mean automating the forecasting of each base-level driver using the most appropriate statistical technique before allowing managers to review and amend the results based on their local knowledge. Or it could mean following the 80/20 rule by automating the forecast for the majority of customers. This can give managers time to concentrate on a small number of large accounts that generate most of the revenue.
• Build collaborative processes using a connected planning platform such as Anaplan that allow all parts of the business to both share assumptions and update their data in real time. That means involving sales and operations (rather than just the finance team) as part of the forecasting process. Only with sales and operations involvement in the forecasting process can the various business functions realign their plans and capacity around changing levels of demand.
• Strive to improve the accuracy of the forecast of each of the business drivers that have the biggest impact on the accuracy of the overall financial forecast. Track accuracy over time and focus initiatives to improve the accuracy of parts of the forecasting model that fall behind.
• Integrate forecasting with management practices so that new forecasts fit into existing cycles for sales or supply chain planning. That way, forecasts will be available when they are most needed and will help managers make better decisions.

Mitch finished his presentation by reminding the audience that companies need to be agile if they are to prosper today. The good news is that agility can be built by systematically adopting the forecasting practices outlined above.