Bayer relies on analytical methods and artificial intelligence (AI) to predict key financial and sales figures. The increasing dynamics of the markets make it difficult for many companies to predict developments and generate a precise sales and financial forecast. AI is supposed to improve this - and provide a reliable look into the future.

In 2015, Bayer began the data.one project to centralise the company's data in order to harmonise control dimensions and provide uniform group reporting. This now enables the group to use AI.

Together with [at] - Alexander Thamm developed Forecasting solution is initially focussing on the further development of purely statistical models based on historical values towards an AI-based model that takes other internal and external data sources into account in particular. In this context, the aim is to simplify the planning of future costs and profits in various business areas. Not only investment decisions depend on these analyses, but ultimately the entire strategic direction. The main aim is therefore to achieve greater integration between operational and financial planning.

AI-based forecasting enables employees to focus on planning strategic products or activities.

Until now, the evaluation has been a time-consuming and highly manual process. With the extensive automation of these processes, these forecasts will no longer be available only at certain planning times. Continuous availability will result in faster responsiveness and decision-making for the company's management. This transformation of the planning process is essential, as the challenges of the Corona pandemic show, for example.

Data science models as a scalable tool for more efficient budget and sales planning

The solution: a data-based forecasting tool based on machine learning. The jointly developed cost forecast is already being used productively and is available to cost centre managers and controllers throughout the group. Based on this, the technological basis for future simplification of the planning process as well as flexible, demand-oriented and continuous forecasting was laid for one business unit as an example. With this pilot, the shift from pure point forecasts to interval forecasts was developed. The latter specify a range of values within which certain events occur with a certain probability. In the future, the solution will make it possible to simulate certain measures, which will serve as a basis for discussion and decision-making. The forecast results are presented in a user-friendly way in an individually designed dashboard. 

"Improved and accelerated forecasts are a valuable addition to the financial toolbox. Forecasting solutions are therefore one of the core tasks and challenges that we want to make available to our finance department in particular as 'IT & Digital Transformation'," says Philipp Plank, Head of Decision Science for Enabling Functions in the Group.

While optimism is justified, it is important to note that these solutions will not replace planning altogether. Rather, they offer the possibility of giving time back to employees by automating much of the forecasting. This allows employees to focus on planning essential, strategic products or activities. "In various projects within and outside the finance function, we have made great progress in the area of automated forecasts through the use of the models and will use them in the future to strengthen the basis for decision-making, for example for production planning and investments," confirms Philipp Plank.

Disclaimer:

This article originally appeared in the Handelsblatt Journal "Künstliche Intelligenz - AI Experience" (June 2022).