Imagine an executive team gathering to discuss its most urgent strategic priority: artificial intelligence. The CEO wants AI to become a new engine of growth. The CIO wants to deploy copilots. The COO sees opportunities to automate processes. The head of product wants AI-enabled offerings. The CHRO is worried about how AI will reshape work. Everyone agrees that AI is the future. Everyone agrees that the company needs to innovate. And everyone leaves the meeting talking about something different.
This is increasingly the real innovation problem inside large organizations. We use one word, “innovation,” to describe fundamentally different kinds of work. Adopting a technology. Improving a process. Launching a product. Creating a business model. Transforming an organization. Responding to a crisis. Everyone uses the same word and assumes alignment. There isn’t any.
I’ve been called the “Dean of Innovation” for nearly four decades, although the title started more as a bit of good-natured ribbing than a credential. I was 27 and Vice President of New Ventures at Domino’s Pizza during the company’s explosive growth in the 1980s when I attended a retreat with Tom Peters, the legendary business author. Somehow “Dean of Innovation” emerged from the gathering, and my older colleagues at Domino’s found it amusing enough to make it stick. Since then, I’ve worked with much of the Fortune 500, the U.S. military, and more cultural organizations than I can remember. After decades of tours of duty inside organizations trying to make innovation actually happen, I’ve become convinced of one thing: almost nobody means the same thing when they use the word.
The result is predictable: resources scatter, expectations conflict, and initiatives are managed with the wrong processes and measured by the wrong metrics. Most companies don’t need more innovation. They need a better way to talk about it.
Stop Asking, ‘Is It Innovative?‘
For decades, executives and academics have tried to define innovation: incremental or radical, sustaining or disruptive, product or process. A better question is simpler: Innovative compared with what? Innovation is positive deviance from the norm that creates value. The same practice can be revolutionary in one company, incremental in another, and obsolete in a third. For a traditional company with decades of legacy systems, using AI to redesign a core process may represent a major departure from the norm. For an AI-native startup, it may be table stakes. The technology may be similar. The starting points are not. That is why leaders should ask two questions: What norm are we breaking? What new value are we creating, and for whom? Without those answers, “innovation” is little more than a corporate aspiration.
Nvidia’s extraordinary rise, for example, cannot be explained simply by saying the company innovated. Its graphics processing capabilities were repeatedly extended into new arenas, most consequentially AI computing. That is the real work of innovation: departing from the existing norm in a way that creates new value. And the farther an idea departs from the norm, the harder it becomes to prove that it will work.
The Data Trap
Executives are trained to demand evidence before committing resources. That is generally good management. It can also kill innovation. The paradox is simple: The more genuinely novel an idea is, the less reliable the historical data becomes. A modest improvement to an existing product can usually be modeled. Customers are known. Costs can be estimated. Competitors can be studied. But what happens when a company enters a market that does not yet exist or deploys a technology whose capabilities and economics are changing every few months?
Consider generative AI. In late 2022, no executive could produce a credible five-year ROI model. The technology, costs, competitors, regulations, and use cases were changing too quickly. The companies that waited for certainty did not reduce uncertainty. They simply learned more slowly. This is the data trap. The ideas with the strongest evidence are usually those most similar to what the organization already does. Demand too much proof, and you unintentionally select for the familiar. Innovation gradually becomes optimization.
Instead of asking, “Can you prove this will work?” ask, “What is the cheapest, fastest experiment that will tell us something important?” In established operations, data informs action. In innovation, action often creates the data. Experiments do not validate the plan. They create the knowledge from which a plan can eventually be built.
Innovation Is Not One Game
Organizations also make the mistake of treating innovation as a single activity. It isn’t. UPS improving route efficiency is largely an optimization challenge. The system exists, data is abundant, and progress can be measured precisely. A company responding to a cyberattack is playing a different game. Speed matters more than perfect information. A legacy company building an AI-enabled service faces uncertainty about customers, pricing, and the business model. Small experiments and staged investments are appropriate. An automaker shifting toward electric and software-defined vehicles faces something larger still: a transformation affecting manufacturing, suppliers, talent, capital, and organizational identity.
All four require good management. They do not require the same management. Yet companies routinely force them through the same stage gates, funding processes, and metrics. Before choosing the process, leaders should ask: How large is the departure from what we already know? How quickly must we move? How much uncertainty exists? The answers should determine the team, funding, governance, metrics, and pace, not the other way around.
Most Corporate ‘Innovation’ Is Actually Something Else
Companies call almost every form of change “innovation”. Installing a new ERP system. Adopting Microsoft Copilot. Reorganizing a business unit. Digitizing a process. Running a hackathon. These activities may be valuable. But they are not the same kind of work. A useful distinction is simple: Creativity produces possibilities. Innovation develops and tests novel possibilities to create value. Change gets people to adopt and scale what has been chosen. Buying an AI platform is technology acquisition. Getting thousands of employees to use it is change management. Discovering a fundamentally new way to create customer value with it is innovation. A company can succeed at the first two without accomplishing the third.
This is why counting AI pilots tells us little about whether an organization is becoming more innovative. Automating an existing process may increase productivity. Deploying copilots may help employees work faster. Those may be excellent investments. But adoption is not invention, and efficiency is not necessarily innovation.
The Danger of False Alignment
Return to the executive meeting. The CFO believes the AI initiative should reduce costs within 12 months. The head of product wants new sources of revenue. The CIO wants a secure technology platform. The CHRO wants to redesign work. All four may be right. But unless those differences are made explicit, the initiative will eventually be judged against conflicting expectations. The organization does not have an execution problem. It has a language problem.
Leaders often respond by pushing harder for alignment. But in uncertain situations, disagreement is information. One executive may see an efficiency opportunity. Another sees a competitive threat. A third sees a new business model. The goal should not be to eliminate those differences too quickly. It should be to make them visible. That is not dysfunction. That is clarity.
A Five-Question Language for Innovation
Executives do not need another elaborate innovation taxonomy. They need a few questions they can use in a meeting, funding review, or conversation with a team. Before approving any innovation initiative, ask five.
1. What norm are we breaking?
Is it a norm inside our company, our industry, the technology, or customer expectations? If you cannot identify the norm being broken, you may simply be improving what already exists.
2. What new value are we creating, and for whom?
Innovation is not novelty. “We need an AI strategy” is no more useful than saying “we need an internet strategy” was 25 years ago. The technology is not the strategy. The question is what becomes possible now that was not possible before. If the value cannot be named, the initiative may be technology theater.
3. How big is the departure from what we already know?
An incremental improvement may involve familiar customers, technologies, and economics. A new business model may call all three into question. Small departures can often be planned. Large departures need to be learned. Match the size of the investment to the amount of knowledge available.
4. How fast do we need to move?
Speed is not always a virtue. A cybersecurity breach may demand action in hours. A new business model may take years to mature. The question is not, “How can we move faster?” It is: What is the appropriate speed for this kind of uncertainty?
5. What is the next experiment?
Not: What is the five-year plan? Not: When can we scale it? What is the next experiment? A good experiment answers an important question at an acceptable cost. Will customers use it? Can the technology work under real conditions? Will someone pay for it? What assumption, if wrong, would cause the idea to fail? The purpose is not to prove that the team was right. It is to learn what is true. The real failure is spending two years building something that could have been disproven in two weeks.
Change the Questions, Change the Conversation
Instead of asking, “Is it innovative?” ask, “What norm are we breaking?” Instead of asking, “What is the ROI?” ask, “What must we learn before investing more?” Instead of asking, “What is the plan?” ask, “What is the next experiment?” Instead of asking, “Are we aligned?” ask, “Where do we see the problem differently?” Instead of asking, “How quickly can we scale?” ask, “Have we created enough value to scale yet?”
These questions do not eliminate uncertainty. They make it manageable. An optimization effort should be accountable for results. An experiment should be accountable for learning. A crisis response should be accountable for speed. A transformation should be accountable for building capabilities. When leaders use the same language and metrics for all four, they create confusion. When they distinguish among them, they can manage each more intelligently.
Innovation Is Something You Navigate
As AI accelerates technological change, more initiatives will be labeled innovative. More companies will launch pilots, labs, and strategic initiatives. More money will be committed before leaders agree on what kind of problem they are actually trying to solve. The companies that succeed will not necessarily be those with the most ideas. They will be the ones that can distinguish optimization from invention, adoption from experimentation, evidence from assumptions, and plans from learning. They will know when to demand data and when to create it. When to move quickly and when to be patient. When to scale and when to experiment.
Innovation has always been difficult to define because it is not a fixed thing. It changes with the organization, the industry, the technology, and the moment. That is why the search for a perfect definition has always been a dead end. Innovation does not need another definition. It needs a language leaders can actually use.
Written by Jeff DeGraff for Fortune as “I’m the ‘dean of innovation’ and I know that nobody can agree what ‘innovation’ actually means” and republished legally through syndication agreement with Reuters Connect. Jeff DeGraff is Clinical Professor of Management and Organizations at the University of Michigan’s Ross School of Business since 1990, DeGraff has established a comprehensive school of thought around systematic innovation development that bridges academic rigor with practical application across business, government, and mission-driven sectors. DeGraff co-created the Competing Values Framework alongside Robert Quinn, John Rohrbaugh, and Kim Cameron — one of the most widely used organizational effectiveness models globally. After earning his PhD at age 25, he served as Vice President of Communications and New Ventures at Domino’s Pizza (1985-1990), where he applied emerging organizational theories to help scale the company from $50 million to $2 billion — earning him the nickname “Dean of Innovation.”
