Is predictive analytics really the answer to your cash flow forecasting (CFF) challenges? It is according to many articles and posts on the subject. They often describe how treasury work treasury workstations and other new applications use predictive analytics to improve short term CFF. What’s not as clear from these articles is how predictive analytics-based applications can:
- Integrate short and long term CFF
- Integrate CFF with foreign currency exposure and working capital forecasting
- Embed CFF into one rolling forecast process that integrates financial and operational forecasts
- Support one scenario planning process that optimizes cash flow, profits and other objective functions
Predictive vs Prescriptive Analytics
The reality is that predictive analytics is not the best tool to support these planning and forecasting objectives. In global manufacturers, for example, none of these objectives can be supported by predictive analytics alone. However, what can support them is prescriptive analytics. The nature of these analytical tools, and the difference between them, are summarized in Exhibit 1 and explained in further detail below.
Predictive analytics is the practice of extracting information from historical and transactional data in order to determine patterns and predict future trends. While demand forecasting is a classic application of predictive analytics, its capabilities go well beyond this. Rather than forecasting aggregate demand for a product, it can identify when a specific customer or customer type will buy a particular product. Other applications of predictive analytics include providing insight into the following:
- Accounts Receivable: DSO for customers
- Manufacturing: machine break downs
- Warranties: claims for product failures
- Marketing: product & customer promotions.
Relative to CFF, what predictive analytics does is improve the accuracy of model parameters, like those shown above. What it doesn’t do is to improve the CFF model itself. This is the domain of Prescriptive Analytics, a formal definition of which are tools that quantify the best outcome among different choices.
When applied to CFF and financial planning and forecasting in general, Prescriptive Analytics provides the means to maintain an operationally realistic model of a business. These tools use mathematical modelling techniques (eg. optimization, simulation) to answer two interrelated questions. What should we do and what are the risks? It includes the type of logic shown in exhibit 2.
In simpler terms, prescriptive analytics can be thought of as a mature form of driver-based planning – a financial planning approach for predicting financial results based on projections of operational volumes. Its modelling logic supports more mature financial planning and forecasting models than do traditional cash flow and financial planning and analysis tools.
While many factors separate the maturity of these models, one is important to understand. Prescriptive analytics support flow-based models that connect products sold to customers, with the resources required to deliver them, across the entire enterprise. This means that this one planning event would automatically quantify the financial and operational resources required for “downstream activities” like shipping, invoicing, storage, transportation, manufacturing, raw material storage, inbound logistics and purchasing.
The use of flow-based approaches is what makes “driver-models” dynamic. They enable upstream (volume and mix) changes to be automatically translated to all downstream activities. They do this by modelling individual connections between activities in the model, for all product and customer combinations. As a result, planning models can result in millions of these flows. What’s also important to appreciate about these models is that:
- Financial and operational resources are defined for all activities in each flow
- Balance sheet and cash flow behaviour is defined for all of these activities
- Financial and operations constraints are defined for all of these activities
- Business rules are defined to resolve these constraints
This modelling logic supports vastly superior scenario planning and cash flow forecasting processes because financial and operational forecasts remain accurate across a wide spectrum of supply and demand scenarios. What’s more, they provide the means to optimize resource allocation by more effectively coping with interconnected trade-offs that govern enterprise performance.
The absence of such logic (from traditional planning tools) is also one of the primary reasons why finance executives struggle to improve forecast accuracy, cash flow forecasting, rolling forecasts and cost and profitability management in global organizations.
The key takeaway is that financial executives need to understand predictive and prescriptive analytics and their relevance to their planning and forecasting challenges. They also need to understand how to effectively embed them into financial planning processes, which is the subject of a separate article which can be accessed here.
Like this item? Get our Weekly Update newsletter. Subscribe today