Innovationsmaschine Data & AI: Warum Produktideen einen harten Wettbewerb brauchen

The Data and AI Innovation Engine: Why Product Ideas Need Real Competition

In the tech world, innovation is one of the most used and most abused terms in corporate communication. Either we are talking about innovation management, a methodically well-developed field that almost no company seriously puts into practice, or we are talking about a buzzword that fills marketing slides and "Our Vision" pages without ever touching the reality of day-to-day business.

And yet right now would be exactly the moment to look more closely.

The quarterly rhythm of the AI revolution

In the tech world, we are seeing innovation leaps in increasingly shorter cycles. Foundation models like GPT, diffusion models like Stable Diffusion, more powerful GPU architectures, and now agentic AI as the next stage of autonomous systems that no longer just respond but plan and act independently. According to Gartner, AI agents and AI-ready data are among the fastest-moving technologies in the current hype cycle. McKinsey reports that 88 percent of companies already use AI in at least one business function, but only around a third are beginning to roll it out at any meaningful scale. The half-life of AI knowledge has shortened from years to months. A CIO recently put it bluntly: the time it takes to evaluate a new technology already exceeds the window in which that technology remains relevant.

New opportunities emerge every quarter, but so does new confusion and overload. It pays to resist the temptation of chasing the first promising idea that comes along and instead stay the course with a clear head and consistent strategy.

The real problem: the governance vacuum

If only it were not so expensive. But the real cost comes from poorly thought-out investments in ideas that nobody seriously questioned, ideas that emerge in a governance vacuum as personal pet projects, reactions to tech hype, or things that landed on the priority list by chance.

Big Tech can afford to crash-land with the Metaverse, Google Glass, or Hyperloop. Most companies and organizations want their ideas to actually grow the business, and they do not have billions in reserve as a safety net.

Despite average GenAI investments of nearly two million dollars in 2024, fewer than 30 percent of AI decision-makers believed their CEOs were satisfied with the return on investment. Another State of AI report shows that only 39 percent of respondents report a measurable EBIT impact at the company level, even though almost everyone is using AI somewhere.

Ideas need real competition

The most productive countermeasure is also the most uncomfortable one: ideas need to compete against each other. Not in the style of Shark Tank entertainment, but through systematic, methodically grounded selection. Which idea has genuine strategic relevance? Which delivers the highest short-term ROI? Which can realistically be executed with the resources available?

And equally important: which ones should be stopped before they turn into zombie projects, kept alive through stubbornness or political consideration long after they have lost any real potential.

Classical innovation management developed a proven tool for exactly this challenge: the Stage-Gate process developed by Robert G. Cooper. It divides development initiatives into clearly defined phases called Stages, each separated by decision points called Gates. At each Gate, an interdisciplinary committee makes a clear call: proceed, adjust, or kill. More than 3,000 companies worldwide use this framework, from P&G and Siemens to LEGO.

What this means for data and AI

The Stage-Gate process is a powerful tool, but it was built for a world of physical products. Data and AI-driven products follow a different logic. They are inherently iterative, heavily dependent on data availability and quality, technically complex, and need to scale alongside the day-to-day business simultaneously.

What is needed is a combination: the methodical rigor of classical innovation management applied to the complexity and layered nature of data and AI-driven products.

I have brought exactly that together in a Data and AI product development matrix that has already been successfully integrated into enterprise operating models. The key elements are:

Role-independent action fields, meaning no silos but cross-functional responsibilities that operate across hierarchies and business areas. Concrete fields include analytics, project management, and consulting. The critical point is that assignment follows capability and interest, not job description. A data scientist can move into project management if that is where they contribute most. A business analyst can take on analytical tasks without officially becoming a data scientist. This breaks open one of the biggest hidden cost traps in organizations: rigid role assignment that prevents available potential from flowing to where it is actually needed. People can almost always do more than their role suggests. It also prevents individual employees from becoming bottlenecks that block an idea from moving forward.

A Stage-Gate process tailored to data and AI, with decision points that evaluate not just technical feasibility but also data availability, ROI potential, and strategic fit. Across the phases from ideation through maturation and evaluation to MVP and product launch, this produces around 40 clearly defined activities. Some examples include collecting and assessing relevant data assets, running ideation workshops, producing effort estimates, developing business cases, and after launch, systematically analyzing and evaluating product performance. Each activity is assigned to an action field and tied to a concrete phase objective. This creates transparency about what is expected from whom and when, and makes visible where bottlenecks are forming or ideas are losing momentum.

A consistent team setup where the people who were involved from the beginning and genuinely care about the idea stay engaged throughout the entire development process.

What the rollout requires and what it cannot tolerate

Experience from enterprise projects makes one thing clear: the matrix does not work by decree. Rolling it out requires a careful touch.

First, it requires genuine management commitment, not lip service but an active willingness to make decisions based on evidence and to own unpopular decisions to stop a project. Without that, any methodology dissolves in the political reality of corporate life.

Second, it requires an honest inventory of existing processes and structures. The matrix is not a replacement for everything that already exists. It is a framework that integrates and builds on what is there. Existing governance structures, reporting lines, and role definitions need to be adapted, not ignored.

Third, and this is the factor most consistently underestimated, it requires experienced consultants embedded from the beginning, ideally in a staff function. Not as external contractors who disappear after the project ends, but as internal navigators. They help steer the process, make the right connections within the organization, support the development of product ideas, keep the innovation portfolio current, and prepare results for C-level steering committees. Especially in the early phase, when ideas are still fragile and teams are still learning how the matrix works, this support is the difference between a living innovation pipeline and another well-intentioned initiative that ends up in a drawer after six months.

What the matrix actually delivers

Experience from enterprise projects shows what this structured approach makes possible. Genuinely innovative ideas using data and AI are generated systematically rather than waiting on coincidence or individual initiative. High-potential candidates with strong ROI are identified and prioritized early, before too many resources flow into less promising directions. The resource problem in demanding day-to-day operations becomes manageable because clear prioritization forces decisions that are often politically difficult to make. And zombie projects with high maintenance costs get shut down as soon as the evidence points against them.

The real lesson

Innovation is not a buzzword and not an end in itself. At a time when AI investment is exploding globally, 37 billion dollars in 2025 alone according to Menlo Ventures, and the gaps between technological leaps keep getting shorter, systematic selection capability is the decisive competitive advantage.

Winning does not go to whoever talks loudest about innovation. It goes to whoever brings the right ideas to market maturity with the right people at the right time.

Anyone who wants to replace a governance vacuum with an innovation engine does not need a new strategy presentation. They need a process that institutionalizes genuine competition between ideas and includes the courage to stop.

What does the innovation portfolio in your organization look like, and who is responsible for deciding what continues and what gets cut? If you sense there is room for improvement, feel free to reach out directly.

References

Back to blog