Not only can organizations leverage data science and machine learning for things like time savings, more efficient processes, and cost optimization, but they can also use it for fully automated cash flow forecasting that can produce results precise enough for the modern enterprise and a changing environment.
What if you could develop automated cash flow forecasting that was 20% more accurate than manual forecasting and that reduced the workload on financial experts by 80%? Later in this blog, we illustrate how one company was able to do just that.
A survey conducted by the European Association of Corporate Treasurers (EACT) in March 2020 revealed that over half of the finance departments surveyed rate cash flow forecasting as a top priority for the next 12 to 24 months. You may ask, though, if forecasts during times of economic change and disruption are inaccurate, should you still bother with cash flow forecasting? The answer is undoubtedly yes — read on to find out how this is possible.
A Real-Life Example
Recently, JTI was recognized by the European Association of Corporate Treasurers (EACT) for its data science project on cash flow forecasting that was performed in Dataiku’s collaborative data science and machine learning platform.
Here’s how it worked:
- First, an initial model was created based on previous data and forecasts and success criteria were outlined, along with corresponding metrics for tracking In this first sprint, predictions could be made with the same precision as in the manual forecasts.
- In the second sprint, the JTI team added external values such as unemployment or gross domestic product (GDP) of the respective geographic regions in which the company is active. These were reviewed for possible correlations.
- In the third sprint, internal factors such as the output of a production site were also examined and no significant changes in the reliability of the forecast were uncovered. In the final sprint, the JTI team optimized the forecast models and created five different variants. One of the models proved to be very flexible with regard to changes, as it takes into account the forecasts for the next month as well as the current figures.
With regard to result specifics, the accuracy of the automated cash flow forecasting is 20% higher than that of manual forecasting. The workload of the financial experts responsible for the forecasts fell by 80% (which translates to 320 hours per month), granting them back more time in their day to be used on other important projects. Further, the more precise forecasts ensure more efficient management of the cash flow as a whole. The Dataiku platform, as demonstrated here, enables business leaders to gain visibility into cash flow forecasts prepared by analyst team members, along with potential scenarios to inform decision making ranging from the more mundane to ones that impact high-level, strategic business objectives.