Innovation is novelty that diffuses: a new solution to an unmet need that relieves a limitation people face and that they adopt. Organizations innovate by designing a purposeful system capable of generating and diffusing that novelty, consistently and at scale. (Innovation, defined)
The first prototype should reduce the uncertainty that matters most
In 2011 we established our innovation lab in a coworking space at 37th Street and Seventh Avenue in Manhattan. We wanted to work among startups, close to their pace and their habits. One afternoon Kirsch, a senior engineer on the team, brought me an idea for a children's tablet app. He wanted to build a minimum viable product, and he estimated close to two months of work to code it for the tablet.
His plan would have spent those two months writing code before any child had tried the app. Some of the open questions concerned the engineering: how the tablet would handle the interactions, and how long the code would take to write. Kirsch could estimate that work from experience. Other questions only children could answer. Would a seven-year-old know which parts of the screen to touch? On each screen, what would a child try to do, and would the app respond the way the child expected? Were the buttons large enough, and far enough apart, for hands whose fine motor control is still developing? Smaller and more distant targets take longer to reach and are missed more often, a relationship known as Fitts's law. [1]
I asked what he most needed to learn first. Which questions required code running on a tablet? Which could a child answer with paper and a pencil? How soon could we put something in front of a child? Each question narrowed the work. Together they pointed to the fastest experiment that would tell us the most.
We started sketching. In about two hours, we had a paper prototype of the interactions he wanted to explore. My boss had three children between seven and ten years old and agreed to bring them to the office the next morning. By lunch, we had our first answers: which parts of the interaction the children understood, which confused them, and what to build next. Two months of building had become a question answered in a day.
The engineering questions involved risk, and the children's questions involved uncertainty. Risk can be measured, because the odds are knowable. Uncertainty cannot be measured, because there is no basis for calculating the odds. That distinction comes from the economist Frank Knight, writing in 1921. [2]
A minimum viable product puts learning first. As the entrepreneur Eric Ries defines it, a minimum viable product is the least effort needed to collect the most useful validated learning about customers. [3] For the children's questions, paper was enough. The questions about how the tablet would perform, and how long the code would take, still needed code.
Prototypes are often the cheapest way to reduce uncertainty. An old saying puts a picture at a thousand words, and the computer scientist Ben Shneiderman extended it: an interface is worth a thousand pictures. [4] A paper prototype is an interface made cheap enough to throw away.
The harder question is which uncertainty a prototype should reduce first. Kirsch's two-month build would have answered whether we could make the app work on a tablet, a question of feasibility. Our paper prototype answered first whether children would understand the app and want to use it, a question of desirability. If the children could not use the design, two months of working code would have answered the wrong question first.
That morning was part of innovating. Whether the app would become an innovation remained open. A team can learn from something that never enters ordinary use. It can also produce something people use without learning what to improve or what to make next.
For a leader, producing an innovation and developing an organization capable of innovating are different achievements. Both draw on the same people and budgets, and they require different work. The first brings something new into practice. The second changes what the organization can notice, make, judge and support.
Innovation brings a possibility into practice
An idea can be original, persuasive and fully funded and still be only an idea. It becomes an innovation when someone puts it into practice and people take it up.
Economic development, in the economist Joseph Schumpeter's account, comes from putting new combinations into practice. His examples included new goods, new methods of production, new markets, new sources of supply and new ways of organizing an industry. A company can innovate through how it serves, supplies or organizes work as well as through what it sells. [5]
The official statistical definition asks for less. The Oslo Manual, the guide the Organisation for Economic Co-operation and Development (OECD) and Eurostat publish for collecting data on innovation, counts a new product as an innovation once it is made available to potential users, whether or not anyone buys it. [6] MITRE's Innovation Toolkit asks for "novelty with impact." [7] Defining innovation as novelty that diffuses adds a further condition: the new thing has to relieve a limitation people face, and people have to adopt it.
Adoption is where leaders can begin to examine value. Who can now do something useful that was difficult or impossible before? What makes the change worth sustaining? What does it cost, and who bears that cost? Adoption gives leaders those questions to investigate. It does not answer them.
Our eBiblioFile service at The Library Corporation began with a specific problem. Libraries needed high-quality catalog records for electronic resources quickly enough for patrons to find newly available material. We asked library discussion lists whether others had the same problem, then interviewed the people who responded.
The team hand-coded and delivered the first batch of records within two weeks of starting the project. Within a year, the service was serving more than 300 libraries. The early work delivered something libraries needed while giving the team a way to learn how to provide it, and we improved the service as libraries used it. [8]
The children helped us investigate an interaction. The libraries received records they could use in their work. Both taught us something. Delivering a service also committed us to libraries in ways the paper prototype never did. The paper prototype was part of innovating. The eBiblioFile service was an innovation: it relieved a limitation libraries shared, and they adopted it.
Usefulness depends on what becomes possible, and for whom
Novelty matters to the people who adopt it when it makes a limitation smaller. "Technology can bring benefit if, and only if, it diminishes a limitation," in the words of the physicist Eliyahu Goldratt, who developed the Theory of Constraints. [9] The benefit can be time saved, lower cost, faster work or something that was not possible before.
"Technology can bring benefit if, and only if, it diminishes a limitation."
Eliyahu Goldratt
The word "can" matters. Diminishing a limitation is necessary for benefit, and it does not guarantee benefit. Whether people take something up also depends on how well it works, what it costs, how much time it asks of them and whether it is socially acceptable.
Applied to innovation, the rule turns usefulness into a question: what can someone now achieve, and what previously stood in the way? Japan's chindogu tradition makes that question harder, and funnier. Chindogu are what Kenji Kawakami called un-useless inventions: each one works, solves a real problem and creates an embarrassing new one. [10] The ingenuity is visible. The reason anyone would keep using one is harder to find.
Design teams judge new ideas with three overlapping questions: is it desirable, is it feasible, and is it viable? The model traces to Larry Keeley of the Doblin Group, whose version named capability, viability and desirability. By October 2003, the designer Peter Merholz could already call the feasibility, viability and desirability Venn diagram "typical." We interpret this as an adaptation of Keeley's model by interaction designers, who replaced "capability" with "feasibility" in those years, and Merholz's post as a record of the change once it had spread. The model's three-part form echoes the Roman architect Vitruvius, whose qualities of good building Henry Wotton rendered in 1624 as "commodity, firmness, and delight." [11]
A chindogu is feasible, because it works. It fails the other two questions: the problem it creates outweighs the one it solves, and nobody makes it for sale.
Timing can change these judgments. The selfie stick appeared in Kawakami's 1995 collection as a joke, and two decades later it had become an everyday object. Diffusion researchers who follow Everett Rogers call the fit between something new and the values, experiences and needs of the people who might adopt it compatibility. [10] An invention's usefulness depends on the circumstances in which people encounter it.
A chindogu raises a further question: does solving the stated problem make the activity better for the person doing it? The selfie stick adds another point: a judgment made in one setting can stop being true as practices and circumstances change.
The costs and benefits of an invention become visible when people use it. Adoption alone cannot establish that a meaningful limitation has diminished, or that the benefits outweigh every cost. Those are judgments about what people can achieve and what the change requires of them.
Making changes what we can understand
Innovating often begins before anyone can state the opportunity clearly. Someone notices a workaround. A new capability changes what a team might attempt. An unexpected result unsettles an explanation that had seemed adequate. There is enough to investigate and too little to specify a finished answer.
Sometimes the use for an invention appears well after the invention. Spencer Silver's weak adhesive at 3M found its purpose when his colleague Art Fry needed a bookmark that would stay in his hymnal, and the Post-it note followed. Biologists call this pattern exaptation: a trait formed for one purpose that comes to serve another. [12]
A hunch becomes a problem a team can state through work that makes the hunch tangible. Prototyping shows what the initial account gets wrong, what existing parts and knowledge can contribute, and what becomes possible at the next scale. This account of innovating comes from Luis Perez-Breva, faculty director of MIT Innovation Teams, in Innovating: A Doer's Manifesto. [13]
A prototype can do more than check whether people like a proposed solution. It can expose a dependency, reveal a mistaken assumption, or help people describe a difficulty they had been unable to explain. Its value depends on what it lets them investigate.
The maker also learns through the act of making. Thought develops through work with the material: the maker notices resistance, adjusts, and discovers something the original intention did not contain. Repeated practice develops judgment about what the material permits and what a particular choice is likely to do. That is the sociologist Richard Sennett's account of craftsmanship. [14]
People adopt a new tool inside an activity they already practice. A hammer belongs to an activity: someone is working with a nail, a board and a purpose. None of those appear in a description of the hammer's properties, and skilled use depends on all of them. When the hammer breaks, goes missing or gets in the way, the nail, the board and the purpose come into view. The essay Heidegger's Hammer develops this idea from the philosopher Martin Heidegger. [14]
The people using what we make already work within such activities. They have habits, obligations, other tools and purposes they share with others. A new service joins an existing practice and changes it. People may use it differently from what its designers expected, combine it with something else, or find that it creates work in a place the team never considered.
Making lets a team meet some of those consequences while there is still time to change the design. A sketch, a manually delivered service, a working component and a technical model each show different things: how materials behave, how people respond, and whether the team's first question was the right one.
The first question may need to change
Every innovation effort begins with a frame: an account of who matters, what needs attention and what would count as improvement. The frame directs the investigation and leaves some things outside it.
A team asked to improve a screen will look closely at the screen. It may miss the policy that forces customers to use it, the information another department already has, or the possibility of removing the interaction altogether. Better execution inside the original frame can leave the source of the difficulty untouched.
In a talk for an innovation lab, I asked people to examine the questions their professions and teams made easy to ask, then the questions they were not supposed to ask. Who sets the boundary? Whose experience disappears from the account? What would someone from another discipline ask first? [15]
A prototype can expose a weakness in the frame, provided the team has permission to change the question. When no result is allowed to change the approved solution, experimentation becomes a way of polishing its presentation.
Strategic foresight asks the same questions about the years ahead: who will matter, what will need attention and what will count as improvement. An emerging technology or a change in what customers expect can reveal an opportunity before anyone experiences a failure. An organization needs ways to investigate possibilities that its current products, customers and measures may give it little reason to notice.
Constraints give inquiry something to work with
Designing a purposeful system for innovation is, in large part, designing its constraints. People need enough structure to make consequential choices together. A team benefits from knowing whom it serves, what it is trying to learn, which commitments it can make, and where the consequences require a wider decision. These conditions let people act without renegotiating everything at every step.
Constraints do different kinds of work. Strategy Is the Making distinguishes five kinds: [16]
- Limiting constraints cap what a stable, well-understood system can produce.
- Enabling constraints let people try different things while their efforts still fit together.
- Governing constraints keep patterns in place with nobody giving an order. Culture, professional norms and platform architecture are common examples.
- Interpretive constraints limit what an organization can perceive, often through the tools and identities that once made it successful.
- Evolutionary constraints appear when a practice that suited one stage of a capability's or technology's development continues into the next stage.
Enabling constraints make coherent activity possible, as the philosopher Alicia Juarrero explains. Shared interfaces, feedback and boundaries can create possibilities that isolated participants would lack. [17] In an organization, a clear decision right can let a team respond while the information is still useful. A limit on concurrent work can protect the attention needed to finish an inquiry.
Other constraints obstruct the work. Access to customers may depend on an approval that arrives after the decision. A funding rule may require a revenue forecast before the team has learned who would buy. A measure designed for a mature service may make an emerging one look unviable by definition.
Leaders need to distinguish what each constraint does. A boundary that protects people from harm, a rule that makes coordination possible, and a limitation that caps delivery call for different responses.
When the aim is to improve an existing system's output, the Theory of Constraints provides a useful discipline: identify the limiting constraint and direct improvement toward it. [9] Making one part faster may achieve little if the work still waits elsewhere. Exploratory work may have to establish what the system could become, and for whom, before its relevant limitations can be understood.
Time requires similar judgment. Some questions can be investigated in an afternoon. Others require repeated use, a seasonal cycle, a technical trial or the slow growth of trust. A short deadline can focus work that fits within it. It can also force a team to report a conclusion before it has learned enough to reach one.
The practical aim is to shorten avoidable waiting while allowing inquiry the time it needs. Funding and review should reflect that distinction. A team should know what it must learn before the next commitment, and leaders should understand why that learning requires the time being requested.
A team's discovery needs a route into the organization
I once worked on a skunkworks team that was given room to explore beyond the established business. Within two weeks, colleagues in product and engineering were calling us "special" and had stopped talking to us. We had also taken six of the company's best platform engineers away from teams that depended on them.
Our freedom had consequences elsewhere. The arrangement concentrated capability in one place and weakened the relationships our work would need in order to spread. Anything we discovered would have to win over the colleagues we had been set apart from.
Exploratory work can need protection. Organizations face a persistent tension between developing what they already know and searching for possibilities whose returns are uncertain, as the organizational theorist James March described. Separate exploratory units can operate under different arrangements while senior leadership keeps them connected to the rest of the company, as research on organizational ambidexterity by the management scholars Charles O'Reilly and Michael Tushman shows. [18] Separation comes with a responsibility to design those connections.
An outside partner, an acquired company or a dedicated lab can contribute knowledge the organization does not have. Whether that knowledge spreads depends largely on how people across the organization work together. Joint research lets colleagues meet the same customer. Shared reviews show where groups interpret the same results differently. Rotations and communities of practice help people develop knowledge that a report alone cannot convey. Decision records preserve why choices changed, so others can examine them after the original participants have left.
These practices need a connection to people with authority. Colleagues may understand what a team learned and still be unable to change the policy, funding or commitments that prevent them from acting on it.
Learning becomes organizational when the system changes: when what a team learned alters a policy, a commitment, an interface, a decision right or a practiced way of working. That is the threshold Strategy Is the Making sets for organizational learning. [16] It is also the point at which a discovery has diffused inside the organization.
A promising practice has to become reliable work
Innovating consistently and at scale means a practice that worked in an experiment has to work in ordinary operations. An experiment can succeed under conditions the rest of the organization cannot reproduce. Its team may have unusually experienced people, selected customers, direct access to a sponsor and permission to bypass established rules. Those conditions can make learning possible. They also make the result difficult to interpret.
Scaling requires a judgment about what produced the result. Direct access to customers may be essential to the new service and need to become ordinary practice. Daily intervention by a senior sponsor may be a temporary support that needs replacing. A policy exemption may reveal a rule that should change, or a responsibility the pilot has yet to learn how to meet.
I once worked with a large furniture retailer whose e-commerce engineering team completed what everyone called a successful agile transformation. Marketing still planned large projects under a different executive. Testing reported to another leader and kept working on its previous cycle. The engineering team sped up, delays grew where its work passed to marketing and to testing, and the promised benefits never arrived.
The handoffs between engineering, marketing and testing limited the result. Changing them meant confronting ownership, priorities and how work passed between functions. The faster engineering practice needed those working relationships to change before it could improve the service as a whole.
A new practice spreads when the relationships, authority, funding and capabilities around it change too. Transition design, developed at Carnegie Mellon University's School of Design by Terry Irwin and colleagues, studies this kind of whole-system change in transitions toward more sustainable futures. [19]
Some existing arrangements must end. A new priority has little force while every previous commitment still draws on the same people. A new decision right remains provisional while an old approval can overturn it. Leaders have to decide what will stop, who is responsible for stopping it, and what must be preserved for customers and colleagues who still depend on it.
The standard for continuing investment should rise as the work matures. Early inquiry should clarify a consequential uncertainty. Further investment should establish whether the result repeats and what it requires. Ordinary operation must deliver reliably, recover when something goes wrong, and meet its obligations without exceptional protection. Counting experiments tells us little about whether those thresholds have been crossed.
Leaders decide what the capability serves
An innovation portfolio expresses choices about the future of the business. Funding core improvements, adjacent opportunities and more uncertain possibilities makes sense when leaders can explain how those commitments support the position they intend to reach. Each commitment consumes attention and capability that cannot be spent elsewhere.
The consequences reach beyond the business. A service can benefit its customers while transferring costs to employees, suppliers or communities. A system can become more efficient while leaving the people who depend on it with fewer choices. Adoption and growth give leaders information about a change. Leaders also have to ask what else the change is doing.
For any significant innovation effort, I want to understand what could become possible, for whom, and what would make us reconsider. The next act of making should help us judge a consequential choice. Before expanding the work, we need to understand which conditions produced the result and whether we can sustain them without passing unacknowledged costs to someone else.
That afternoon with Kirsch began with an uncertain idea and an estimate of two months. By the following day, the children had given us a clearer account of what our design allowed them to do. The next decision was still ours, and so was the responsibility for what came after it.
In an organization capable of innovating, discoveries like that change what gets funded, how colleagues work together and what the organization becomes able to offer. The making changes the thing being made and, when the learning reaches far enough, the organization making it.
If your leadership team is working out how to develop that capability, request a briefing.
Talks and workshops on these ideas
- The One Thing in the Way: a keynote that applies to strategy Goldratt's rule that a technology can bring benefit only when it diminishes a limitation.
- Redesigned to Disrupt: a keynote on designing an organization that finds new value while it runs what already works.
- Why Transformations Stall: a keynote on what happens when ways of working spread without the context that made them succeed.
- Portfolio Strategy, Flow and the Constraint: a workshop where leadership teams decide what to commit, stop or transform.
Questions leaders ask about innovation
What is innovation?
Innovation is novelty that diffuses: a new solution to an unmet need that relieves a limitation people face and that they adopt. Organizations innovate by designing a purposeful system capable of generating and diffusing that novelty, consistently and at scale.
What is the difference between invention and innovation?
An invention is something new that has been made. It becomes an innovation when it relieves a limitation people face and they adopt it. Joseph Schumpeter drew the line between invention and innovation, and the OECD's Oslo Manual makes implementation the defining condition.
Why does adoption matter in the definition of innovation?
Adoption is the point at which novelty starts to change how people work or live. Relieving a limitation is a necessary condition for adoption, and people also weigh how well the new thing works, what it costs, how much time it takes and whether it is socially acceptable. Chindogu inventions from Japan are new, real and working, and they stay on the shelf, because using them costs more than the task is worth.
What are the types of innovation?
Schumpeter named five: new products, new methods of production, new markets, new sources of supply and new ways of organizing an industry. The Oslo Manual groups innovations into two types, product innovations and business process innovations.
Does incremental improvement count as innovation?
Yes, when the improvement is new to the people adopting it and relieves a limitation they face. Small routine adjustments fall below that threshold.
What is the difference between innovation and innovating?
Innovation is the outcome: novelty that relieves a limitation and diffuses. Innovating is the work of generating that novelty and helping it diffuse. A purposeful system designed to repeat that work does it well.
What is the difference between risk and uncertainty in innovation?
Risk can be measured because the odds are knowable. How long a feature will take to build is one example. Uncertainty cannot be measured because there is no basis for calculating the odds. How a child will respond to a new interface is one example. Innovating well usually means finding the uncertain questions first and answering them with the fastest experiment that works.
How do constraints affect innovation?
Constraints do different kinds of work. Limiting constraints cap what a system can produce, and Goldratt's Theory of Constraints aims improvement at the one that caps the whole. Enabling constraints, such as decision rights and limits on work in progress, let people try different things while their efforts still fit together. Governing constraints keep patterns in place. Interpretive constraints limit what an organization can see. Evolutionary constraints tie practices to stages that have passed.
Sources
- Paul M. Fitts, "The Information Capacity of the Human Motor System in Controlling the Amplitude of Movement," Journal of Experimental Psychology 47, no. 6 (1954): 381 to 391. https://doi.org/10.1037/h0055392
- Frank H. Knight, Risk, Uncertainty and Profit (Houghton Mifflin, 1921), part III, chapters VII and VIII. https://www.econlib.org/library/Knight/knRUP.html
- Eric Ries, "What Is an MVP?"; and The Lean Startup (Crown Business, 2011). https://leanstartup.co/what-is-an-mvp/
- Ben Shneiderman and Catherine Plaisant, Designing the User Interface: Strategies for Effective Human-Computer Interaction, 4th edition (Pearson Addison-Wesley, 2005), section 14.5, "Information Visualization."
- Joseph A. Schumpeter, The Theory of Economic Development, translated by Redvers Opie (Harvard University Press, 1934), chapter II; and Business Cycles (McGraw-Hill, 1939).
- OECD and Eurostat, Oslo Manual 2018: Guidelines for Collecting, Reporting and Using Data on Innovation, 4th edition (OECD Publishing, 2018), chapters 2 and 3. https://doi.org/10.1787/9789264304604-en
- MITRE Innovation Toolkit, "What Is Innovation?", 10 February 2020. https://itk.mitre.org/2020/02/10/what-is-innovation
- Simon Marcus and William Haas Evans, American Library Association presentation (2013), eBiblioFile case study, slides 33 to 35.
- Eliyahu M. Goldratt, "The Reason for Technology," Necessary and Sufficient series, session 1 (Goldratt Group). https://toc-goldratt.com/en/product/n-and-s-series-1-the-reason-for-technology; see also Eliyahu M. Goldratt, Eli Schragenheim and Carol A. Ptak, Necessary but Not Sufficient (North River Press, 2000).
- Kenji Kawakami, 101 Unuseless Japanese Inventions: The Art of Chindogu, translated by Dan Papia (W. W. Norton, 1995); Everett M. Rogers, Diffusion of Innovations, 5th edition (Free Press, 2003), chapter 6.
- Alan Cooper, The Inmates Are Running the Asylum (Sams, 1999), chapter 5; Hugh Dubberly, "Alan Cooper and the Goal Directed Design Process," Gain: AIGA Journal of Design for the Network Economy 1, no. 2 (2001). https://www.dubberly.com/?p=155; Peter Merholz, peterme.com, 21 October 2003. https://peterme.com/archives/week_2003_10_19.html; Vitruvius, De architectura 1.3.2, translated by Morris Hicky Morgan as The Ten Books on Architecture (Harvard University Press, 1914); Henry Wotton, The Elements of Architecture (London, 1624).
- Lemelson-MIT Program, "Art Fry and Spencer Silver: Post-it Notes." https://lemelson.mit.edu/resources/art-fry-spencer-silver; Stephen Jay Gould and Elisabeth S. Vrba, "Exaptation: A Missing Term in the Science of Form," Paleobiology 8, no. 1 (1982): 4 to 15. https://doi.org/10.1017/S0094837300004310
- Luis Perez-Breva, Innovating: A Doer's Manifesto for Starting from a Hunch, Prototyping Problems, Scaling Up, and Learning to Be Productively Wrong (MIT Press, 2017), especially chapter 2. https://mitpress.mit.edu/9780262536127/innovating/
- Richard Sennett, The Craftsman (Yale University Press, 2008); Martin Heidegger, Being and Time, translated by John Macquarrie and Edward Robinson (Harper & Row, 1962), sections 15 and 16; Hubert L. Dreyfus, Being-in-the-World: A Commentary on Heidegger's Being and Time, Division I (MIT Press, 1991); and William Haas Evans, "Heidegger's Hammer," Fugue Strategy Advisors. fuguestrategy.com/essay-heideggers-hammer.html
- William Haas Evans, How Frame Exploration Enables Innovation, Nxt Lab opening presentation (spring 2022), slides 10 to 17.
- William Haas Evans, Strategy Is the Making: How Organizations Shape What Becomes Possible (forthcoming). https://sitm.foresightops.org/book
- Alicia Juarrero, Context Changes Everything: How Constraints Create Coherence (MIT Press, 2023); William Haas Evans, "Introduction to Design Studio Method," LinkedIn, 20 February 2014. https://www.linkedin.com/pulse/20140220213016-13520960-introduction-to-design-studio-method/
- James G. March, "Exploration and Exploitation in Organizational Learning," Organization Science 2, no. 1 (1991): 71 to 87; Charles A. O'Reilly III and Michael L. Tushman, "The Ambidextrous Organization," Harvard Business Review 82, no. 4 (April 2004): 74 to 81.
- Terry Irwin, Cameron Tonkinwise and Gideon Kossoff, "Transition Design: An Educational Framework for Advancing the Study and Design of Sustainable Transitions," 6th International Sustainability Transitions Conference, University of Sussex (2015).

