There isn’t a one-size-fits-all approach to budgeting or forecasting. Some organisations keep it relatively high-level using a subset of accounts, while others conduct planning at the same levels of actuals. There is one common denominator, however: speed.
Regardless of what organisation I have consulted or worked in, management want forecast numbers as quickly as possible (with the exception of zero-based budgeting, but perhaps I’ll cover that another time). It helps them make decisions for the company in the short and long term. The sooner they can do this, the better. Business analysts need to get data into an EPM system quickly yet accurately. There are varying approaches to this such as:
- using a prior forecast as a starting point
- using top-down allocations
- using centralised drivers that populate P&L accounts for each period
There is nothing wrong with these approaches or using a combination of methods. They all have their merit and I advocate for them. However, there is a little-known yet potentially powerful feature called Predictive Planning that can be used to quickly populate a forecast.
How it works
It has often been said that to look at the future, we have to look at the past. This is exactly what predictive planning does. The prior periods of an individual account, entity, cost centre, etc., are analysed and a prediction is formed for the next 12, 24 or even more months, depending on how much historical data in available for the prediction. The EPM solution selects (or you can choose, in some solutions) an appropriate time series algorithm, detects any seasonality and produces the data. The prediction also shows the margin of error with a best case and worst case scenario that potentially could occur based on the prior periods used. This is also visually represented in a chart showing the historical actuals and then future periods. Let’s take a look at this in action.
Show and tell
For this example, I am using Oracle PBCS which has this predictive planning functionality out of the box. I have created a data form with actuals for the prior year for an entity and cost centre. I am going to use this data to populate the forecast scenario and the version “predictably” called prediction.
I do the following:
- Select Predictive Planning from the data form option
- The prediction is calculated and the data produced for the accounts for the entity and cost centre
- Review the data
The prediction is visualised in a chart. I can select any account that has actuals and view the prediction. I can click on any line on the chart and see the figures and the period. The actuals are plotted together with the subsequent prediction. Note that if historical periods are missing, this data is also predicted using the same algorithm.
The other information instantly available is:
- Prediction method (this can differ per account depending on the historical data)
- Confirmation of the date ranges predicted
- If Seasonality was detected
The raw output data can also be viewed and the forecast periods predicted examined more closer within the chart.
Copy to a scenario and version
The true benefit is that the predicted data can be copied and pasted to a scenario and version combination. It is – without exaggeration – possible to produce forecasted figures for the entire P&L for both revenue and costs in minutes (yes, you read that correctly).
The paste options allow you to choose from the prediction data but also the worst or best prediction figures. The period range and whether all the members, accounts in this case, are to be populated.
The Paste button is pressed and then the magic happens.
It would of course be tempting to leave it there but we all know that cannot be the reality. The predictive planning functionality gives you a starting point. It is explainable and gives you a feel for the trajectory of the business based on historical performance. Predictive Planning does not of course take into account future plans of the business, such as planned new hires, investments or a sales initiatives.
Historical trends might also not actually be realistic. I have a retail customer who had to close a number of stores during the pandemic. This has skewed the historical trends, so it isn’t the case of treating the predictive plan as the golden standard to which the actual forecast should aim to mimic. There is still a huge need for a human element to increase the accuracy of the forecast. The results need to be correctly interpreted with business analysts and controllers acting as a business partner to guide the business through the decision-making process.
A recommended use case is to use predictive planning as a starting point or a reference point to allow comparison as a sense check. It complements the forecast process, allowing more time for value-adding processes such as “what if” scenario planning or helping to produce a 12-month rolling forecast as actuals are updated each month.
So there you have it. A quick, logical and explainable way to get the forecast started using predictive planning. Just remember to add your tweaks.