Why Successful Data & AI Strategies Start with the Business Problem
The most important question didn’t come up until an hour later
A few months ago, we facilitated a workshop on artificial intelligence. The objective was clear: to work together to identify the potential AI offers the company and determine which use cases to start with. The discussion started as expected. We talked about automation, new tools, and the question of where AI can help save time, streamline processes, and enable better business decisions. The participants contributed many ideas. Before long, a substantial list of potential applications had emerged. Then someone asked a question that changed the entire workshop:
“What problem are we actually trying to solve?”
The question was neither particularly complicated nor surprising. Nevertheless, it caused a moment of silence. Suddenly, everyone realized that they had been talking about solutions for quite some time without having clearly identified the underlying problem. From that moment on, the focus was hardly on AI at all. Instead, the conversation turned to slow decision-making processes, coordination between departments, a lack of transparency, and processes that had evolved over the years and now tend to hinder rather than support the company. In retrospect, that was precisely the most valuable part of the workshop—not the discussion about AI, but the realization of where the company is actually losing time, speed, and room to maneuver.
Why Many AI Initiatives Start in the Wrong Place
This situation is by no means unusual. In fact, we see it all the time. It’s not because companies are asking the wrong questions. The real challenge is that the discussion often starts with the technology. As soon as the topic of AI comes up, attention automatically turns to models, tools, and potential use cases. And that’s understandable. Anyone who attends industry events, reads trade publications, or scrolls through LinkedIn today is constantly bombarded with new success stories. New tools and supposed breakthroughs appear practically every week. The pressure to take action grows accordingly. That’s exactly why many companies initially focus on the question of where AI can be applied.
From a strategic perspective, however, another question is far more important: Where is the company actually losing value today?
After all, AI doesn’t generate value simply because it’s being used. Value is created only when a specific problem is solved—when decisions can be made faster, when employees spend less time on routine tasks, when knowledge becomes more readily available, or when customer inquiries are handled more efficiently. Those who start directly with the technology run the risk of compiling an impressive list of possible applications without knowing which ones are actually relevant.
Many companies are looking for use cases, even though they should actually be looking for bottlenecks
Interestingly, the number of possible AI applications is rarely the problem today. In workshops, we often hear statements like, “We’ve already gathered 30 or 40 ideas.” This is generally a positive sign; it shows that the organization is engaging with the topic. At the same time, however, it is precisely this multitude of possibilities that often leads to a fallacy. Many companies view the number of identified use cases as a sign of progress. In reality, however, it says little about whether the relevant problems are being addressed.
Here’s an example: A company develops an AI assistant that provides internal information more quickly. Technically, the solution works excellently. However, usage figures fall short of expectations.
Upon closer inspection, it becomes clear that the real problem wasn’t the search for information at all. Employees often didn’t know which information was authoritative. Different systems provided different answers. The challenge lay less in finding knowledge than in its quality and authority.
The use case made sense. The bottleneck was simply located elsewhere.
This is precisely why companies should not start by asking which AI solutions are conceivable. They should first understand which obstacles are currently preventing certain goals from being achieved.
The Real Task Is Not to Implement AI
From our perspective, a successful data and AI strategy is often misunderstood. The task is not to implement as many AI applications as possible. The task is to eliminate operational bottlenecks. This difference may seem minor at first glance, but it fundamentally changes the approach. For example, if a company regularly loses valuable time on manual coordination tasks, AI is not the starting point. The starting point is the question of why these coordination tasks are necessary in the first place. Only then can one assess whether data, automation, or AI can be part of the solution. The same applies to knowledge management, customer service, product development, or internal processes. Technology should always be selected based on the problem—not the other way around.
The most successful AI initiatives we see therefore usually begin with a very objective analysis: Where are delays occurring today? Which decisions take too long? Which activities tie up capacity unnecessarily? Where are costs arising that could actually be avoided? Those who answer these questions clearly usually realize surprisingly quickly which use cases take priority and which do not.
Conclusion: The Most Important AI Question Isn’t an AI Question
When companies talk about AI, the discussion often revolves around tools, models, and technologies. That’s understandable. After all, these topics are visible and tangible. The real challenge, however, usually lies elsewhere. Companies rarely gain an advantage by identifying the most AI applications. They gain an advantage by solving the right problems.
That’s why a good Data & AI strategy doesn’t start with the question, “Which AI should we use?” but with the far more important question: “What’s preventing us from being more successful today?”
Those who find a clear answer to this question make better decisions about data, processes, and technology. And that’s exactly when real added value is created. Because AI isn’t a goal in itself. The goal is better decisions, better processes, and better results.
