Predictive Planning is a key trend towards a data driven forecast. For sure it is not applicable for each and every situation as the data might not hold all the relevant variables and not everything is predictable. It is however a key element to scale the organization as where it is applicable it can automate a great deal of the forecasting activities and even improve accuracy (Roche example).
The data driven approach is for sure not new to the CoE central forecast team in SAP global controlling, where we have used a data-first approach to successfully challenge the usual bottom-up and improve accuracy in the process for the last few years. Furthermore we have leveraged SAP Analytics Cloud Predictive Planning capabilities to forecast our Travel expenses (blog).
With success however comes the burden of an ever-increasing list of tasks to be completed, and thus the search for further automation continues. We have now looked at making the process of automating expense account forecasting with Predictive Planning replicable. Predictive Planning in a box if you will.
I have teamed up with the CoE central forecast team to generate such a process looking at multiple accounts with seasoned controllers. In this blog I'd like to share the main learnings and outcome of the process.
Automation was the mantra throughout, eliminating as many repetitive tasks as possible. Using the Multi-action framework we could now automate: Data load, running predictive forecasts, allocations and direct writeback to planning versions. Learn all about multi actions here.
(Continuous) Check for fit
We found that some expense items have gone through tremendous change recently. Changing Travel expenses during the pandemic as an intuitive example. With these changes, recent data to train the predictive models was not suitable. Here we now advise to keep running the automated model side by side with the manual process, thus we can track when the automated approach delivers the results needed. This works in reverse as well, when automated models no longer provide the accuracy desired, manual approaches might be preferrable again which brings us to the next point.
Check for accuracy
Past results are never an assurance for the future; therefore, it is advisable to analyze performance over time of a variety of models. Firstly, for accuracy and secondly for understanding of the business trends driving the data and as such the forecast.
Make it pretty and understandable
Knowledge is useless unless shared, hence the use of commenting to add a narrative was found very useful to explain the choice of which model and thought process surrounding the choice. Beyond the narrative we found sharing the results with accompanying visuals boosted the desired interaction and conversation, so yes appearances do matter! Luckily this has been made easy with embedded formatting and Business Intelligence capabilities in SAP Analytics Cloud.
Next steps
With these findings in place and a repeatable process with accompanying models & reports created in SAP Analytics Cloud, the central forecast team is now self-sufficient to apply it to all the expense accounts and over time roll-out the capabilities to the overall Profit & Loss Accounts.
My counterpart in global controlling Eric Schünemann has gotten so excited and self-sufficient with SAP Analytics Cloud, that he has written about the experience from his perspective which I’d recommend as further reading here. He made sure my support became quite “hands-off” as he was able to become an SAP Analytics Cloud master in no-time.
The data driven approach is for sure not new to the CoE central forecast team in SAP global controlling, where we have used a data-first approach to successfully challenge the usual bottom-up and improve accuracy in the process for the last few years. Furthermore we have leveraged SAP Analytics Cloud Predictive Planning capabilities to forecast our Travel expenses (blog).
With success however comes the burden of an ever-increasing list of tasks to be completed, and thus the search for further automation continues. We have now looked at making the process of automating expense account forecasting with Predictive Planning replicable. Predictive Planning in a box if you will.
I have teamed up with the CoE central forecast team to generate such a process looking at multiple accounts with seasoned controllers. In this blog I'd like to share the main learnings and outcome of the process.
Automation was the mantra throughout, eliminating as many repetitive tasks as possible. Using the Multi-action framework we could now automate: Data load, running predictive forecasts, allocations and direct writeback to planning versions. Learn all about multi actions here.
(Continuous) Check for fit
We found that some expense items have gone through tremendous change recently. Changing Travel expenses during the pandemic as an intuitive example. With these changes, recent data to train the predictive models was not suitable. Here we now advise to keep running the automated model side by side with the manual process, thus we can track when the automated approach delivers the results needed. This works in reverse as well, when automated models no longer provide the accuracy desired, manual approaches might be preferrable again which brings us to the next point.
Check for accuracy
Past results are never an assurance for the future; therefore, it is advisable to analyze performance over time of a variety of models. Firstly, for accuracy and secondly for understanding of the business trends driving the data and as such the forecast.
Make it pretty and understandable
Knowledge is useless unless shared, hence the use of commenting to add a narrative was found very useful to explain the choice of which model and thought process surrounding the choice. Beyond the narrative we found sharing the results with accompanying visuals boosted the desired interaction and conversation, so yes appearances do matter! Luckily this has been made easy with embedded formatting and Business Intelligence capabilities in SAP Analytics Cloud.
Next steps
With these findings in place and a repeatable process with accompanying models & reports created in SAP Analytics Cloud, the central forecast team is now self-sufficient to apply it to all the expense accounts and over time roll-out the capabilities to the overall Profit & Loss Accounts.
My counterpart in global controlling Eric Schünemann has gotten so excited and self-sufficient with SAP Analytics Cloud, that he has written about the experience from his perspective which I’d recommend as further reading here. He made sure my support became quite “hands-off” as he was able to become an SAP Analytics Cloud master in no-time.
- SAP Managed Tags:
