Quality of Work Life: an evolving definition
Automated forecasting, instant reporting and optimized decision-making, that's what Algorithmic Forecasting promises tomorrow's Finance Department.
In the past ten years, digital transformation, and Big Data in particular, has already changed the way companies and their CFO Office work. Among the opportunities offered by Big Data, Machine Learning is one of the most attractive technologies today. According to Market Research Future (Machine Learning - Global Forecast to 2024), the Machine Learning market, estimated today at $ 7.3 billion, will grow by 42.8% per year over the 2018-2024 period to reach $ 30.6 billion in 2024. Like the Sales / Marketing and Supply Chain departments, Finance departments must now grasp the subject and anticipate its challenges and opportunities, in particular those offered by "Algorithmic Forecasting".
Machine Learning is an AI technology capable of improving its own activity independently from available and structured data. For example, when you browse on Amazon, Machine Learning intervenes to predict the evolution of your future needs according to your research and thus offer you increasingly appropriate suggestions. Algorithms, on the other hand, are machine-driven tools that perform predefined and purposeful tasks. Algorithmic Forecasting combines these two tools. This can now benefit the CFO Office and management controllers, by improving performance management, developing forecasts and detecting new growth drivers.
First, Algorithmic Forecasting generates significant time savings by automating the classification, construction and analysis of data. Its use can reduce reporting work by up to 50%. The algorithm is based on data processed by Machine Learning and collected by the Cloud to instantly generate a report. Additionally, according to the study conducted by the "Journal of Business Forecasting" of The Institute of Business Forecasting & Planning, the implementation of this solution in various companies has increased forecast accuracy from 5% to 15%, reaching accuracy ranging from 85% to 95%, without any human interventions. In short, the algorithm generates a financial report faster, more precisely and without potential human bias.
The savings in time and accuracy generated by Algorithmic Forecasting are highly beneficial for financial management. On one hand, by freeing up some time of the management controller, they allow more exhaustive and relevant financial analyses. On the other hand, these financial reports reach Senior Management more quickly, allowing more time for decision-making. More advanced than Enterprise Performance Management (EPM) tools alone, Algorithmic Forecasting is becoming an operational, strategic and decision-making tool.
For example, thanks to its "Merlin" Algorithmic Forecasting tool, Daimler has many algorithms specially modeled to predict its different KPIs. Coupled with Data Visualization, the tool makes it possible to present the results in a relevant and efficient way to controllers.
Indeed, Algorithmic Forecasting does not impose forecasts, but offers a plurality of different scenarios based on variables defined by controllers. Consequently, the application and the use of Algorithmic Forecasting allow to prioritize and refocus the activities of the CFO Office around more value-added tasks, such as identifying new growth drivers or optimizing processes.
According to a study by the French National Association of Financial Directors and Management Control, more than 60% of companies want to optimize their budget process. Today, the idea of an automated budget still raises many challenges, both on the organizational and technical aspects.
And yet, the beginnings of budget automation are already there. Using its "Merlin" tool, Daimler is thus able to predict the next 18 months with more than 100 different KPIs in more than 50 entities, ultimately preparing the group’s half-yearly and annual plans automatically.
This is also the case with Uber, which has gradually turned to "automated budget". Initially, the company implemented a Rolling Forecast covering 18 months to replace its traditional budget. Then, Uber perfected and automated its forecasts thanks to their solution, "Michel Angelo", from Algorithmic Forecasting in PaaS (Platform as a Service).
"Michel Angelo" trains with Data training and increases his forecasting capacities almost autonomously. For example, the best predictive models in the business are capable of calculating more than 250,000 forecasts per second to automatically determine the best scenario. Already used at Uber, the combination of Rolling Forecast and Algorithmic Forecasting seems to be the most viable solution in the automated budget.
Primarily used to date by large companies, the development of Algorithmic Forecasting in small and medium-sized companies is imminent. According to the CIO Survey 2019 conducted by Gartner, 1 in 10 companies were already using more than 10 AI applications in 2019 (chatbots, Process Automation, etc.). In the future, with technological advancement, Algorithmic Forecasting will complete this portfolio by allowing total automation of the budget.However, the implementation of such a solution is not easy and requires certain prerequisites and good practices which must be anticipated.
Data Science and Machine Learning techniques have not yet found their ground within Finance Departments. However, after years of incremental optimizations within the Finance Function, covering process re-engineering, and Shared Services optimization, etc. Finance Departments have an opportunity to drastically improve their business partnering capabilities.
Investments into R&D models will obviously take some time, to identify the right drivers, their impacts and their interdependencies. Yet, the ability to create automatically baseline forecasts in a stable environment, while facilitating decision-making and performance management through eased impact modelling is definitely appealing enough to start investing.
It generally starts with proof-of-concepts in some specific areas of the business (demand forecasting, external risks & opportunities forecasting, etc.) that help grasp the key concepts and potentials, before industrializing the method.