Executive Summary
The conventional advice for innovation teams is clear: define the problem first, then solve it. But a study of over 1,100 real innovation teams tells a different story. Dr Johnathan Cromwell, Associate Professor at the University of San Francisco, shares research showing that teams which embrace ambiguity and let the problem emerge through the process of doing consistently outperform those that lock down their problem definition too early.
Core insights:
- Teams that started with low problem clarity but gained it by the midpoint achieved an implementation rate of over 80%, versus approximately 50% for teams that started with high clarity and maintained it
- Problem and solution are not sequential steps — they co-evolve, each shaping the other in a continuous dance
- The midpoint of any project is a critical inflection point: missing it without a coherent problem-solution match should be treated as an alarm bell
- Early divergent thinking followed by convergence is the mechanism through which unclear problems become clear
- AI is not just solving old problems faster — it is revealing problems we did not know existed, making this new mode of thinking increasingly essential
- The VP of Innovation’s job is to resist dismissing projects that lack a clear problem definition early on — that ambiguity may be a signal of future breakthrough, not failure
Rethinking Einstein's Problem-Solving Wisdom
There is a quote widely attributed to Albert Einstein: “If I had an hour to solve a problem, I’d spend 55 minutes defining the problem and five minutes solving it.” It is clean, memorable, and, as we discovered in a previous episode with Zorana Ivcevic Pringle, almost certainly apocryphal.
What Einstein and his collaborator Leopold Infeld actually wrote in 1938 is rather more interesting: “The formulation of a problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill. To raise new questions, new possibilities, to regard old problems from a new angle requires creative imagination.”
Note the shift. Einstein was not prescribing a linear timeline. He was saying that formulating a problem is itself an act of creative imagination — not a prerequisite to creativity, but an expression of it. That distinction sits at the heart of Johnathan Cromwell’s research, and it has significant implications for how business leaders manage innovation teams.
The Dance Between Problem and Solution
Most innovation frameworks present the process as a sequence: understand the problem, then generate solutions. Academic literature, stage gate processes, and consulting methodologies all tend to place the problem firmly in the lead.
Johnathan’s research challenges this assumption — not by dismissing the importance of problem clarity, but by questioning whether the problem must always be the starting point.
“Think of it like a dance,” he explains. “One partner leads, the other responds. In problem solving, either the problem or the solution can take the lead in shaping the other. The traditional assumption is that the problem leads. What we tested is whether that is always the case.”
The answer, it turns out, is no.
What 1,100 Innovation Teams Revealed
In “The Hidden Power of Messy Teams,” published in MIT Sloan Management Review in spring 2026, Johnathan Cromwell and Jean-François Harvey studied an innovation competition run by a Fortune Global 500 company, which by the time they encountered it had grown to more than 1,100 participating teams from around the world. Crucially, this was not simply a winner-takes-all contest. Around 650 teams followed through to implement their innovation in the business — generating additional revenue and savings totalling more than $60 million — giving the researchers a rich dataset covering the full spectrum of outcomes, not just the winners.
The team measured two things at the start and at the exact midpoint of the competition: how clearly teams understood the problem they were solving, and how much divergent and convergent thinking they were doing.
The headline finding overturns received wisdom.
Teams that started with very clear problem definitions and maintained that clarity throughout achieved a project implementation rate of approximately 50% — slightly below the competition average. Teams that started with lower clarity — more ambiguity, less shared understanding of what they were solving — but gained clarity by the midpoint achieved an implementation rate of over 80%.
“This jump in performance leaves a big question,” Johnathan says. “Why are teams that are clearer about their goals less successful at actually delivering results?”
Why Early Clarity Can Constrain Innovation
The answer lies in degrees of freedom.
When a team locks down its problem definition early, it constrains the solution search to what that problem permits. This feels productive — milestones are clearer, progress is easier to measure, good ideas are easier to evaluate against a fixed criterion. But that constraint cuts off the territory where breakthrough solutions live.
The teams that lived with ambiguity longer were free to explore both the problem and the solution simultaneously. As they evaluated competing ideas, they were forced to surface their own priorities and assumptions — and through that process of convergence, the problem became clear not as a precondition of the work, but as an outcome of it.
“Early divergence, later convergence, even though the problem isn’t known, can be a mechanism to find the problem,” Johnathan explains.
This reframes ambiguity not as a risk to be managed but as a resource to be exploited.
The paper illustrates this vividly with two contrasting cases from the competition. One team defined its problem with high precision from day one — a billing dispute issue affecting 60% of customer complaints — and was selected early by managers as a highly promising project. It ultimately failed to get implemented due to unexpected hurdles when scaling the solution. A second team began with a clarity rating in the bottom 10% of all teams, exploring a hazardous field practice for technicians. Drawing inspiration from a retired engineer neighbour, a local flea market, and a mountaineering documentary, they eventually converged on a solution that not only improved safety, speed, and reliability but also sharpened their understanding of the problem itself. It was judged one of the top ten projects in the entire competition.
The Midpoint Matters More Than the Starting Line
Johnathan draws on teams research from the early 1990s to explain why the midpoint of a project is so critical. Across projects of all durations — from one hour to one year — teams naturally pause at the halfway point to reflect on what they have done and begin planning what comes next. This is not a management intervention; it is a natural human dynamic.
The implication for innovation portfolio managers is significant. The midpoint is the right moment to ask whether a team has found a coherent match between problem and solution — not the starting gate.
“If you float through the midpoint of a project and you still haven’t found that coherent match, that should be an alarm bell,” Johnathan says. “Your job as a leader is to facilitate that process — to understand it as a natural dynamic, not to panic, and to help others not panic.”
The three phases he describes are open exploration, convergence on a problem-solution match, and then focused execution. Miss the transition into phase three, and the execution phase never gains traction. The research is unambiguous on the cost of missing it: teams that failed to converge by the halfway point saw their implementation rate plummet to just 25%.
Why This Challenges the Stage Gate Orthodoxy
The stage gate process exists for good reasons. Innovation budgets are finite, and portfolio managers need a disciplined way to kill projects early so that the genuine winners receive sufficient funding. The desirable-viable-feasible framework gives teams clear hurdles to clear.
But Johnathan’s research suggests that applying these hurdles too rigidly, too early, disadvantages exactly the kind of messy, exploratory projects that tend to produce the highest-impact innovations.
“The end point of any innovation is to find a cohesive match between problem and solution,” he argues. “If you anchor the problem too early, you miss out on degrees of freedom. But you can also constrain the solution and keep the problem open — that gives you just as many possibilities.”
A useful illustration comes from manufacturing. American automotive manufacturers traditionally handed tier-one suppliers rigid specifications — exact dimensions, narrow tolerances. Japanese manufacturers gave their supplier partners rough boundaries and asked for the optimal solution within them. The latter approach consistently produced better outcomes, precisely because it preserved degrees of freedom on both sides of the problem-solution equation.
The message for VP-level innovation leaders is pointed: do not dismiss projects that lack a clear problem definition early on. Build a different set of metrics for different points in the process — and resist the temptation to treat early ambiguity as evidence of a team that has not done its thinking.
Jobs to Be Done: Useful, But Not the Whole Story
The conversation turns to Jobs to Be Done, one of the most compelling frameworks in the innovation toolkit. The principle is elegant: instead of asking what people want, ask what outcome they are trying to achieve, independent of any particular solution. Henry Ford captured the idea intuitively — people do not want a faster horse, they want to get from A to B.
Johnathan respects the framework but pushes back on one of its foundational claims: that Jobs to Be Done are stable over time.
“I would imagine that as technology evolves and as we have new experiences and capabilities, the core of what problems need to be to be effective innovations in the Jobs to Be Done framework probably changes over time,” he says. “Our needs today are fundamentally different from what they were five years ago — let alone fifteen.”
The more important discipline, he suggests, is not finding a permanently stable job but understanding the underlying driving force behind why a goal matters to the customer. This is where the five whys technique earns its place: each successive why peels back a layer of assumed solution space to reveal a deeper motivation. Ask why someone wants to get from A to B, and the answer changes the solution space entirely — one person needs to be physically present for a friend’s wedding; another only needs to communicate with their bank manager and can do so by phone. The same surface-level job, the same opening why, but two entirely different answers two or three levels down — and two entirely different innovation opportunities as a result.
How AI Is Expanding the Problem Space Itself
Johnathan’s Harvard Business Review article on AI and innovation identifies three distinct modes in which companies are responding to the technology.
The first is exploiting existing problems. Instacart, for example, used AI to deliver groceries faster and more conveniently — the same problem, done better.
The second is expanding the problem. Khan Academy moved from passive self-guided learning to proactive personalised education, then extended its AI-powered tools to teachers for lesson planning — a significant expansion of scope, still within the same broad domain of education.
The third — and most powerful — is exploring entirely new problems. These are problems customers did not know they had, and would never have articulated if asked. The Steve Jobs quote that captures this comes from when he was asked whether Apple did any market research for the iMac: “It’s really hard to design products by focus groups. A lot of times, people don’t know what they want until you show it to them.”
AI-powered personal assistants for non-executives illustrate this third mode. If you had asked people beforehand what their biggest problems were, they would have listed scheduling inefficiencies and competing demands — problems they were trying to solve with calendar tools. They would not have said “I need a comprehensive personal assistant.” The solution revealed the problem.
As Mark observed during the conversation: our problem space is constrained by our known solution space. AI expands the solution space dramatically — and in doing so, it exposes problems that previously lay beyond the horizon of what seemed solvable. Organisations that explore new technologies with an open mind about what problems they might reveal will find problems their competitors have not yet imagined addressing.
A Different Mental Skill Set — and One Worth Developing
This mode of emergent problem solving is harder than conventional divergent thinking, Johnathan notes. His research shows it is approximately 30% more difficult for people than the traditional approach of knowing the problem and generating multiple ways to solve it.
The reason is that inferring a problem from something concrete and tangible — a technology, a capability, a product — requires holding uncertainty in a way that feels uncomfortable. The brain wants to converge. Resisting that pull long enough to let the problem surface is a learnable skill, but it takes deliberate practice.
Johnathan has developed a personality assessment for creative thinking that identifies individual preferences — whether someone naturally leans towards divergent exploration, convergent evaluation, or emergent thinking. Understanding those preferences at the team level allows leaders to design processes that play to the full range of cognitive styles, rather than assuming everyone approaches problems in the same way.
What the VP of Innovation Should Do Differently on Monday Morning
Johnathan’s practical advice is directed squarely at the portfolio manager under pressure to show return on investment.
“Encourage and accommodate projects that have not yet figured out their core Job to Be Done. Create a new set of metrics at a different point in the process to evaluate whether a project has the true potential to go the distance — not just whether it has a clear problem at the outset. Look at the relative clarity of the problem at the midpoint, not the starting gate. And do not dismiss projects simply because they cannot yet articulate a sharp problem definition.”
That ambiguity, handled well, may be precisely what makes them worth funding.
Listen to the Full Conversation
Dr Johnathan Cromwell and Mark Blackwell explore emergent problem solving, the hidden power of messy teams, Jobs to Be Done, and what AI is doing to the innovation problem space in full in this episode of Arkaro Insights.
📺 Watch on YouTube: https://youtu.be/1xV8HarXli8
🎧 Listen on Buzzsprout: https://www.buzzsprout.com/2012667/episodes/19263901
About the Speakers
Dr Johnathan Cromwell is Associate Professor of Entrepreneurship and Innovation at the University of San Francisco School of Management, where he also serves as Faculty Director of the Entrepreneurship and Innovation Initiative. His research focuses on creativity, innovation, and teams, particularly when working on vague, open-ended, and ambiguous problems — such as the application of AI and other emerging technologies to different industries and use cases. His work has been published in Administrative Science Quarterly, Research Policy, Harvard Business Review, and MIT Sloan Management Review, and has won multiple Best Paper awards at leading academic conferences. He holds an S.B. in Chemical-Biological Engineering from MIT and a Doctorate in Management from Harvard Business School.
Connect with Johnathan on LinkedIn: www.linkedin.com/in/johncromwell/
Mark Blackwell is founder of Arkaro, specialising in change management, innovation, and commercial excellence for the agriculture, food, and chemicals industries. Arkaro’s collaborative “do it with you” approach works closely with clients through a four-step process: Understand, Co-create, Enable, Sustain. We don’t just coach — we get on the pitch with you.
Build Innovation Processes That Work With Complexity, Not Against It
If your organisation is struggling to generate breakthrough innovation despite smart people and disciplined processes, the issue may not be execution. It may be that the process itself is constraining the problem space before the team has had the chance to find it.
At Arkaro, we work with organisations in agriculture, food, and chemicals to build innovation capability that is genuinely adaptive — processes that embrace the right kind of ambiguity at the right moment, and convert it into clarity and implementation. We don’t just share frameworks; we work alongside your teams to embed them.
Contact Mark Blackwell: mark@arkaro.com Connect on LinkedIn: www.linkedin.com/in/markrblackwell
Related Reading from Arkaro
Guests mentioned in this episode
Why Creative Problem-Solving Teams Select the Wrong Ideas — Dr Roni Reiter-Palmon on the 53% finding: the most creative teams spend over half their time framing the problem before generating solutions. The direct precursor to Johnathan’s research on what happens when that framing unfolds through doing rather than upfront analysis.
AI vs Human Creativity: Why Creativity Is a Choice, Not a Gift — Dr Zorana Ivcevic Pringle on the Einstein quote we unpick in this episode’s opening, and why the formulation of a problem is itself a creative act. She is also the researcher who confirmed that Einstein’s famous 55-minute quote was apocryphal.
AI and Jobs to Be Done: Where Human Judgement Wins — Scott Burleson on the Jobs to Be Done framework discussed throughout this episode, and specifically where AI can and cannot assist in the innovation process.
Jobs to Be Done: The Missing Link in B2B Innovation — Scott Burleson’s pyramid framework and why B2B purchasing decisions involve emotional and identity considerations as much as functional needs.
On disruption and why companies miss it
Why Smart Companies Miss Disruption: The 3 Ghosts Blocking Innovation — Scott Anthony on the organisational patterns that prevent established companies from responding to disruption — the companion piece to Johnathan’s argument that locking down the problem too early forfeits exactly the degrees of freedom you need.
On creativity, constraints, and AI
Creative Constraints: Why Less Freedom Yields More Innovation — Dr Catrinel Tromp on how boundaries shape creative thinking, connecting directly to Johnathan’s finding that ambiguity is a resource, not a risk.
Slow AI: Why Rushing to Solutions Kills Creativity — Dr Vlad Glaveanu on using AI to generate better questions rather than quick answers — a natural extension of the emergent problem-solving mode Johnathan describes.
On adaptive strategy and complexity
Emergent Strategy: Why Five-Year Plans Fail — Pete Compo on why strategy itself must be emergent rather than planned, echoing Johnathan’s finding that the most valuable innovation outcomes cannot be fully specified in advance.
B2B Innovation Strategy: Avoiding the Quick Win Trap — How short-term portfolio pressure crowds out the exploratory work that produces breakthrough innovation.
Reference Material
- Cromwell, J.R. & Harvey, J.F. (2026). The Hidden Power of Messy Teams. MIT Sloan Management Review, Spring 2026. The practitioner article at the heart of this episode — and the source of its title. Argues that teams which discover the problem over time consistently outperform those that define it at the outset.
- Cromwell, J.R. & Harvey, J.F. (2025). A Problem Half-Solved Is a Problem Well-Stated: Increasing the Rate of Innovation Through Team Problem Discovery. Research Policy, 54(3). The full academic study underpinning the SMR article, tracking problem clarity and implementation rates across 1,100 innovation teams at a Fortune Global 500 company.
- Harvey, J.F., Cromwell, J.R., Johnson, K.J. & Edmondson, A.C. (2023). The Dynamics of Team Learning: Harmony and Rhythm in Teamwork Arrangements for Innovation. Administrative Science Quarterly, 68(3). The paper connecting team learning patterns to innovation outcomes, co-authored with Amy Edmondson of Harvard Business School.
- Harvey, J.F., Cromwell, J.R., Johnson, K.J. & Edmondson, A.C. (2025). New Research on the Link Between Learning and Innovation. Harvard Business Review. The practitioner companion to the Administrative Science Quarterly paper.
