Predictive Analytics im Forecasting – Status quo in österreichischen Unternehmen

Abstract

Due to a more dynamic and uncertain business environment, forecasts are becoming more difficult to predict. Therefore, traditional planning and budgeting processes which are based on rigid assumptions, reaching their limits, as they can only respond to changes to a limited extent. Digitalization opens up opportunities in forecasting particularly through the use of predictive analytics. Serveral practical examples and a wide range of literature sources give the impression that predictive analytics is already widely used in forecasting. However, current studies show that the use of such a technologie remains low. Since these studies primarily focus on the German market, a research gap arises concerning the current status in Austrian companies. Therefore, this master thesis examines the existing challenges, barriers, success factors, practical recommendations and application areas for Austrian companies in the use of predictive analytics in forecasting. Furthermore, this study aims to identify reasons why some companies decide not to use predictive analytics in forecasting. For this reason, this master thesis investigates the status quo of predictive analytics in forecasting within Austrian companies using a systematic literature review following Fink. Furthermore a qualitative study based on Mayring is used in the empirical part of this thesis. Additionally, a category system was developed from the analysis of 30 relevant publications, which formed the basis for data evaluation. In total 15 interviews were conducted and analyzed with a total of 17 experts. The results of the study show that Austrian companies consider predictive analytics beneficial for revenue forecasting and cash flow/liquidity planning. Furthermore, the current challenges and barriers for companies by using predictive analytics in forecasting are acceptance and trust issues, as well as a lack of expertise. Technological challenges such as concerns around data protection, data security, and the quality and availability of data also persist. Regarding the reasons why companies decide not to use predictive analytics in forecasting, the results show that scarce financial and human resources, the business model or acceptance and trust issues are the main reasons for not using it. The results of the success factors indicate that that building trust and acceptance, launching pilot projects and establishing cross-functional teams are essential for successful usage. Additionally, the controlling department has a crucial role, as it is expected to proactively lead and promote the usage of predictive analytics in forecasting. Finally, the results show that running predictive forecasts in parallel with traditional forecasts is a crucial recommendation. This means that the predictive forecast should be used as a "second opinion”. This approach promotes trust and acceptance in predictive models. Moreover, it is necessary to ensure the traceability of the forecast models. The findings show that this can be achieved through "Explainable AI”, as this makes it possible to explain and recalculate the calculation path of the predictive forecasts. Finally, if the current data basis is insufficient, data preparation and cleaning are recommended as essential actions.

Date of Award	2025
Original language	German (Austria)
Supervisor	Peter Hofer (Supervisor)