Why do companies often fail to realize the benefits of AI?
Many companies are already using AI, but have not yet achieved measurable business benefits. The key difference usually lies not in the technology itself, but in clear objectives, suitable data, defined responsibilities, and consistent integration into processes and decision-making. Those who want to use AI effectively should not start with as many AI tools as possible, but rather with the right use cases, rapid validation, and a clear understanding of the intended value.
Key points
- High AI usage is not yet proof of economic success.
- AI must be consistently embedded in decisions and processes.
- Added value is created only through clear goals, relevant data, and defined quality requirements.
- Successful initiatives start with a focus, learn early, and do not scale too quickly.
- AI requires leadership, accountability, and transparent guidelines.
Widespread adoption, limited benefits
Artificial intelligence has become established in businesses and is influencing areas such as product development, customer interaction, and internal processes in equal measure. According to the “AI Index Report 2025 by Stanford Human-Centered AI”, approximately 78 percent of companies worldwide use AI, often across multiple areas simultaneously.
This finding gives the impression that AI is already firmly embedded in value creation. In practice, however, the reality is different: despite significant investments in technologies, data platforms, and tools, the economic benefits often fall short of expectations. Productivity gains remain sporadic, decisions are rarely altered, and the desired competitive advantage fails to materialize. The mere introduction and use of AI is therefore not yet a reliable indicator of actual business success.
An analysis by the Boston Consulting Group illustrates that only about five percent of companies use AI in a way that has measurable effects on operational efficiency, growth, or EBIT. In contrast, AI often remains stuck in the experimental stage: Projects may start with clear expectations and show initial results, but then they lose momentum. Technically, the models work, but their results are not consistently incorporated into strategic or operational decisions. With isolated use cases, standalone solutions, and little lasting impact, fragmentation results instead of scaling. For companies, this means that the economic benefits of AI lag behind its technological potential.
Quality matters, not technology
There is a key difference between successful and failed AI initiatives: It is not the AI technology that determines the added value, but the quality of the fundamentals. Companies must ask themselves: Are our goals clearly defined, is our data up-to-date and relevant, and have we defined how AI results should be used in day-to-day operations?
Successful approaches therefore do not aim for as many AI functions as possible, but rather for rapid iterations and early learning. AI should be deployed where it creates concrete added value in process optimization. The roadmap is simple: test early, learn continuously, and improve on an ongoing basis.
AI requires clear leadership, not just technology
For companies, this means that AI is not a traditional IT project. It directly impacts decision-making processes and changes how organizations operate and are managed. To achieve this, AI needs clear guidelines:
- When are AI results reliable?
- Who is responsible for decisions?
- How are uncertainties or errors handled?
Studies on Responsible AI show that the greatest benefits arise when companies consider business goals, data strategy, and quality requirements together from the very beginning. AI realizes its value not through maximum automation, but through transparent, trustworthy support for human decisions.
AI in business: from implementation to measurable impact
Companies can drive AI initiatives on their own, but in practice, important questions often remain unanswered.
- Which use cases actually contribute to the corporate strategy?
- Which data is critical, and which is not?
- Where is it worth starting small rather than scaling up immediately?
Questions like these should be answered especially in the early stages, because AI initiatives begin with orientation and goal-setting rather than with the mere implementation of AI tools. Management, in particular, should have a shared understanding of what AI is supposed to achieve, where its limits lie, and how its success is measured. This prevents short-sighted, reactive action and lays the foundation for the sustainable use of AI.
FROX is happy to support you in conducting a structured assessment if you are planning to establish AI in your company. Together, we create transparency regarding where potential already exists, what prerequisites are still missing, and how AI can be meaningfully integrated into your existing processes. Whether AI is used in ITSM and Customer Service Management (CSM), in Business Process Automation (BPA) or in service management areas such as IT Operations Management (ITOM) and Configuration Management Database (CMDB): We align your AI and business goals.
