AI in controlling: booster for the visibility of value contribution?
With the use of AI and automation, the tasks in controlling are changing noticeably. Routine activities are taken over, analyses are available more quickly and the basis for decision-making is broader or, if necessary, more focused.
The focus of tasks for controllers is shifting. Data preparation and standard analyses are becoming less important. As a result, the value contribution of controlling is also changing. Thanks to technological developments, it is becoming more clearly recognizable, but processes that run in the background are no longer perceived and valued as a controlling service.
This raises key questions for the role of controlling:
- Which tasks will remain with Controlling – and which will be permanently replaced by AI and automation?
- Does the elimination of routine activities actually make the value contribution of controlling more visible – or rather less tangible?
- Are expectations of controllers already changing – away from the preparation of results towards the classification and interpretation of results?
- Which skills become more important when technological systems take over a larger part of the analysis?
- How can the contribution of controlling to corporate management be clearly presented and communicated in the future?
What is your opinion on this, what are your experiences? We look forward to discussing this with you!
Prof. Dr. Christoph Eisl is a professor at the Faculty of Business and Management at the University of Applied Sciences Upper Austria, Campus Steyr, and a member of the ICV Board. He has been working for many years on the further development of controlling and finance functions in the area of conflict between digitalization, management and value contribution.
Contributions to thesis 2
“AI does not automatically make the value contribution visible, but is based on existing data models, KPI definitions and derivation logic. If these are not consistent, nothing becomes clearer through automation, inconsistencies are only reproduced more quickly. Visibility increases. Understandability does not. The bottleneck is not in the automation, but in the derivation of the figures.”
Structural Financial Authority | Mandated by Owners, PE & Boards when figures are no longer explainable
“Dear Holger, thank you very much for your contribution – I can follow your reasoning very well. “Garbage in, garbage out” still applies: If data quality, KPI definitions and derivation logics don’t fit, automation won’t help either. However, AI can help to make inconsistencies in large data sets visible.
In tests with GenAI tools, for example, we have found that incorrect or missing postings and deviations from planned values can already be identified quite well (we have published simple practical examples in Controller Magazin).
If the database is clean, the controlling department has more scope for analysis, interpretation and the development of measures together with management. GenAI tools can also increase productivity, for example through MS Copilot in Excel. However, the additional value contribution does not arise automatically.
In your opinion, what specifically is needed to increase the value contribution of controlling?”
Prof. Dr. Christoph Eisl, Professor at the Faculty of Business and Management at the University of Applied Sciences Upper Austria, Campus Steyr, and ICV Board Member
“Christoph Eisl Thank you. However, your example in the comment is at transaction level. Recognize incorrect postings, identify missing postings and deviations from planned values. All correct. But there is another level between the transaction and the control decision, namely the data path. In other words, everything that turns a raw transaction into a decision-relevant key figure. Aggregation, periodization, KPI logic, planning assumptions.
This is exactly where the breaks occur. However, this is not because bookings are missing, but because the derivation logic between ERP, BI and reporting and whatever else is out there, such as planning, consolidation, forecasting tools, etc., is not consistently defined. AI works on this data path, but it does not clarify it. Value contribution only arises at management level when figures not only stand out, but can be consistently explained from the origin to the key figure. The causal chain in the thesis only works if the data path is already clear. And this is precisely where the gap lies. You defend the transaction level. But the thesis claims control level. The data path in between remains open in both.
And that, in turn, is a structural problem, not an automation problem.”
Structural Financial Authority | Mandated by Owners, PE & Boards when figures are no longer explainable
“Holger Breuer Thank you for your comments. Your points are very easy to understand and I support your argument. I didn’t want to defend anything or praise AI to the skies, but above all to initiate a discussion about the extent to which controllers are given freedom and how they can best fill this freedom in order to increase their own “value contribution” (however this is defined). You have provided valuable input here with your comment.”
Prof. Dr. Christoph Eisl, Professor an der Fakultät für Wirtschaft und Management der FH Oberösterreich, Campus Steyr, sowie ICV-Vorstandsmitglied
“I would like to ask a supplementary question for discussion:
Which skills do you think will become more important in the future if technological systems take over a larger part of the analysis?”
Prof. Dr. Christoph Eisl, Professor at the Faculty of Business and Management at the University of Applied Sciences Upper Austria, Campus Steyr, and ICV Board Member
“I agree. With more automation, the value clearly shifts. Away from creation, towards decisions, ownership and what actually results from it.”
Guido Diaz, Head of FP&A & Controlling | E-Commerce & SaaS Finance Leader | Driving Agile FP&A & Digital Transformation
