Unprecedented technology-driven changes are predicted to create opportunities and threats in all societies and professions over the course of the next decade. This development affects entire organizations and requires individual departments to transform. Among these, the finance department, once a front-runner in terms of technology adoption, is in danger of falling behind.
Different digital technologies have a strong impact on finance processes and the way employees in the finance department work. For instance, proficiency with new information technology and statistical analysis become more and more important for management accountants.

This work contributes to the digitalization of the finance function along three main research questions. In doing so, it follows a design science research in information systems approach and proposes contributions to the knowledge base in several respects. Firstly, the development of a future target state for the finance function and a way of benchmarking an organization against it are proposed.
Secondly, the thesis takes a closer look at machine learning and analytics adoption – two of the currently most promising digital tech¬nologies for finance departments – with the help of a structural equation model. Thirdly, it introduces use cases in financial accounting with the implementation of machine learning for better process efficiency and accuracy in invoice processing and in management accounting for improved cash flow forecasting.

Addressing the first research question, a zero-quartile benchmarking approach is proposed where companies collectively define a best possible target state and benchmark their core finance processes on an activity level against it.
Key findings include (1) automation and analytics provide a benefit for many process activities, sometimes unexpected, (2) a digital enterprise platform is particularly helpful for enterprise performance management, (3) the record-to-report process can be treated with lower priority as it benefits from improvements in upstream processes, and (4) the biggest lever identified is in assessing the credit worthiness of a customer with the help of analytics.
The most important findings relating to the second research question are (1) currently, there is hardly any full-scale adoption, (2) prototypes for different use cases and with different algorithms mostly use structured internal data, (3) none of the constructs in the combined technology-acceptance / task-technology-fit model can be neglected, (4) however, task characteristics currently have the biggest influence, and (5) familiarity with machine learning and time-series analysis is rather low.
Finally, this work provides several guidelines for a more successful implementation of machine learning and analytics in financial and management accounting based on the two prototypes developed.