Advanced analytics for financial planning & analysis (FP&A)
Key considerations in designing next generation FP&A processes
A volatile world puts FP&A in the spotlight
Business stakeholders are making new demands of financial planning and forecasting. They are looking for relevance, timeliness and actionability: business steering requires up-to-date financial plans. The global pandemic exposed the weaknesses of traditional planning processes and forced Finance teams to revert to spreadsheets and financial modelling to provide support for decision-making at a volatile time.
Now the need is to prepare for future volatility by developing scenario modelling and moving towards continuous planning cycles with extended time horizons. In addition, it is essential to link financial and operational plans closely in order to ensure optimal alignment across business functions, especially at volatile times. While the direct effects of the global pandemic are declining, new uncertainty is being generated by geopolitical risks and war, supply chain disruptions and inflation, according to the latest edition of Deloitte’s Swiss CFO survey. How do Finance teams react?
Leading FP&A teams design the next generation planning process
Finance teams realise they need to act. Over 60 % of organisations who participated in Deloitte’s global planning and budgeting survey intend to make changes to the way they carry out FP&A. The changes range from extending planning horizons, increasing the frequency of plan updates or even changing the approach to planning. Leading organisations have already made progress and are better integrating and connecting operational and financial planning. More than 50% of Finance teams have set up well-connected sales and finance plans and 35% have linked up personnel and financial plans. There is further scope to make planning better connected, such as close integration of marketing plans and related performance assessment into financial resource allocation. Planning for sustainability initiatives and their impact on financial indicators, a newly emerging theme, can be expected to gain further traction soon.
We see progress in making planning more connected, but organizations vary in their level of adoption of advanced analytics as a planning method. Only a quarter of Finance teams apply algorithmic forecasting. Among those using it, the predominant approach is to build customised models outside collaborative FP&A planning platforms for selected use cases. Scalability is limited and broad adoption has not yet been fully achieved.
There are successful use cases in sales & demand planning, where predictive methods can generate detailed baseline forecasts in a fraction of the time needed by humans. Advanced planning methods use driver-based models as a basis and tailor them to the underlying business model. Leading external indicators can further improve the forecast accuracy and impact of advanced models.
Modelling the cost of goods sold there have been challenges due to the complexity of product cost accounting and the measurement of underlying value flows and intercompany effects. Successful use cases are still rare in this area. Some organisations successfully extend predictive models into operating expenses and cash positions. Many more organisations could apply advanced analytics techniques and draw benefit from them.
Leverage advanced analytics to generate insights and drive efficiency
Advanced analytical methods can address a variety of use cases within financial planning & analysis. These range from supporting descriptive exploration of underlying developments and their root causes to prescriptive decision-support and recommendation of future courses of action to optimise financial performance. A helpful method is to look at use cases and problem types that can be addressed using advanced methods.
These analysis types help FP&A teams identify potential use cases that can benefit from advanced analytical methods. It is essential to perform a rigorous definition of each use case and to assess the expected value that can be contributed. These value contributions range from reaching a higher level of efficiency through automation of analysis to higher quality decision support by generating actionable insights and recommendations. One use case that addresses both efficiency and process quality is the application of predictive analytics to generate baseline forecasts.
The application of predictive forecasting methods helps FP&A teams to generate forecasts at a lower cost to the organisation. Forecasting frequency can be accelerated to support business steering with up-to-date predictions. In addition, predictive forecasting methods generate unbiased estimates of future outcomes while still allowing for human adjustments to adapt the predicted course of action or integrate effects that the model could not anticipate, such as, for example, a planned restructuring.
Integrate predictive forecasting into your performance management
Early explorers limit the use of predictive forecasting to enriching existing performance reports with system-generated forecasts – for example by replacing run-rates. Showing the predictive forecast next to the human forecast can serve as additional information, allowing predictive methods to be introduced and stakeholders to gain an initial experience without immediate changes to ways of working. Such Report enrichment does not fully move the needle, as forecasts are still generated in traditional ways.
Algorithmic validation integrates predictive forecasts into traditional forecast generation. Predictions can influence the forecast submission; however, the planner fully owns data preparation and submission. The human forecast is the leading outcome. Algorithmic validation is the first type of hybrid forecast.
Process integration can be further advanced by providing planners with prefill optionality as the second form of a hybrid forecast approach. Planners can choose to pre-populate their forecast submission using predictive models, replacing the effort of manual data preparation. Predictive prefilling can be offered side by side with other references, such as the budget or the previous year’s actual data or latest demand planning data. Machine-supported forecasting leverages the benefits of predictive capabilities, while still giving planners the option to choose individually about the most feasible forecast generation approach. Adjustments can still be entered on top of the data prefill.
Machine-generated forecasting puts algorithm in the driving seat, populating predictive baseline forecasts. The machine forecast can be adjusted by human enrichments in defined effect categories. Algorithms and humans populate forecast outcomes in a symbiotic way. While machines generate forecasts based on internal and external input factors, humans focus on analysis of underlying business developments and the interpretation of outcomes and mitigation as well as acceleration measures towards financial targets.
Machine intelligence sees forecasts generated by machines based on algorithmic models without further human enrichments. A fully automated, singular forecast can be generated in high frequencies at low organisational cost to provide near-real time predictions of business developments as soon as updated data is available. While Finance teams still regard human enrichment of forecast data as essential, visionary organisations aim at machine-generated forecasting as the ultimate step.
From our perspective, predictive forecasting is a must-have capability for modern FP&A organisations to steer the business effectively and attract talent. Leveraging machine-generated forecasting with human enrichment in a symbiotic way is the most promising approach today. We expect broad adoption of algorithmic forecasting within 2-4 years.
We would be pleased to help you pave the way towards predictive forecasting in your organisation. Please do not hesitate to reach out to us.
This article has been authored by Jochen Bueter, Deloitte’s offering lead for planning, budgeting and forecasting.