Does Your Company Need an Innovation Lab? It Depends

It depends on five things. First, whether innovation is on the leadership agenda with a sponsor who will show up; without that, a lab is a room. Second, where the company stands on the Innovation Maturity Index: below innovation-engaged, awareness comes first. Third, whether a validated opportunity has somewhere to go, a product team or a venture path. Fourth, how large and divided the company is, which decides between one team and a network of them. Fifth, what you want the lab to do: enable the company to innovate, or innovate on its behalf. The first survives; the second is the lab that gets closed. Most companies that ask the question need the innovation function, and only some of them need a lab.

• The function, not the place: what every company needs is what Innovation Mode 2.0 calls the innovation function, “an always-on system of processes, services, technology tools, knowledge, and talent, aiming to maximize the ability of the organization to discover and pursue opportunities.” A lab is one way to house part of it
• Sponsor: the program “needs executive sponsorship from the top level of the organization.” If the CEO will not stand on a stage for it, do not build it
• Maturity: the level decides the first move; see at what maturity level a lab makes sense
• A path to market: the lab's output is a validated opportunity, and someone has to build it; see what happens when nobody does
• Size and structure: one team with manual services for a mid-size company; a network of innovation cores for a group with divisions; see lab or team and several labs
• The mandate: enable or build. The design guide takes it from here: what the lab's mandate should be

Usually the wrong one. A lab is bought as the cure for a company that feels it is not innovative, which is a symptom, not a diagnosis. Innovation Mode 2.0 traces corporate innovation failure to six deficits: leadership, organizational design, innovation capabilities, real-world connection, talent and culture, and venture building. I read a lab as a capabilities answer. It does nothing for a leadership deficit, where innovation is not on the agenda, and little for a culture deficit, where people do not trust the process. Set up in those conditions, it becomes the company's proof that it innovates, which the book calls innovation theater, and it damages the culture it was meant to lift.

• Diagnose first: the six deficits behind corporate innovation failure and the way to diagnose yours come before any org chart
• What a lab fixes: a capabilities deficit, when the company has the will and the people but no method to capture, assess and validate ideas. Those capabilities can also arrive as services without a lab. The book's warning: “setting up expensive labs or a makerspace featuring fancy gadgets, devices, and tools may be impressive at first sight and add some value, but unless they are strategically positioned, they cannot move the needle toward a more innovative version of the company”
• What it cannot fix: a leadership deficit. In the book's words, “Innovation is not part of the leadership agenda” is the mark of an innovation-inactive company, and no lab moves it there
• What it makes worse: a culture deficit, when people “don't believe in the company's innovation promises” and perceive the effort as “innovation theater.” A showcase lab confirms the suspicion
• The right question: not whether we need a lab, but which deficit is holding us back and what the smallest capability that addresses it would be
• The honest case for a lab: a company with a sponsor, a community of innovators and a flow of problems that has nowhere to test them. That company needs validation capacity, and a lab is a fair way to provide it

The evidence is thinner than the claims. Vendors say labs are the way to innovate, critics say most of them fail, and neither side shows its data. What the record does show: most large companies opened one, far fewer think they have a company-wide culture of innovation, and the labs that closed ended in two ways: some as showrooms nobody asked for, others folded into a business that had learned to do their work. In Innovation Mode 2.0, I describe what many organizations have witnessed: “the failure of stand-alone innovation labs that produced impressive prototypes but minimal business impact.” The labs that work are the ones that made the rest of the company better at innovation; the ones that closed as failures were the ones that innovated instead of it.

• How many have one: 87% of the respondents to a 2017 Capgemini study had an innovation center, and 17% thought they had a company-wide culture of innovation. Two years earlier, 38% of the world's 200 largest companies were experimenting with one. The gap between having a center and having a culture is the whole story
• What companies say when they close one: Walmart closed Store No. 8 in 2024, its incubator since 2017, moving around 300 people into its technology and innovation operations; Walmart Canada integrated Blue Labs into the business in 2023, eight months after its public launch; Nordstrom scaled back its lab in 2015 with the words “Rather than just a team focused on innovation, it's now everyone's job”
• The quiet closures: Disney closed its Boston research lab in 2016, opened in 2011 with six researchers, with the statement “We regularly evaluate our organization to meet the evolving needs of our business.” A 2017 list names six closures between 2013 and 2016, Disney's among them
• What the closures have in common: the function was absorbed; the place was shrunk or dropped. Nordstrom's sentence is the book's mandate in a retailer's words: the team's job is to make innovation everyone's job
• What “failure” means: an OECD OPSI analysis of 137 public innovation labs lists “closing a lab means failure” among its fallacies: “Not every end is a testament of failure, just like not every renewed investment reflects solid work.”
• The claim nobody can source: the “up to 90%” figure; see whether 90% of innovation labs fail

Nobody has shown it. The figure circulates in trade press, consultancy posts and vendor pages, attributed to a 2015 Capgemini report and repeated from page to page without a link. The 2015 report, as the trade press reported it, measured adoption: 38% of the top 200 companies were experimenting with centers. None of the coverage gives a failure rate. A 2017 Capgemini study found that 87% of respondents had a center and 17% thought they had a company-wide culture of innovation, which is a gap, not a failure rate. The earliest sentence I could trace says “up to 90%, one expert said,” with no name and no link. Treat the number as a rumor, and the gap as the fact.

• The sentence: “A 2015 report from Capgemini found that the vast majority of innovation labs – up to 90%, one expert said – fail to deliver on their promise of adding value and generating growth,” in an insurance trade article of 2019, without a link to the report
• What the 2015 report said: 38% of 200 of the world's largest organizations were experimenting with innovation centers, 309 centers counted, as reported when it was published
• What the 2017 report said: 87% of those surveyed had an innovation center; 17% thought they had a company-wide culture of innovation; 76% were still at the building stage, from a survey of 1,700 employees at 340 organizations
• Why it matters: a decision made on a rumor is theater of another kind. The real finding is harder and more useful: most companies have a center, and most do not think they have a company-wide culture of innovation
• What to measure instead: adoption of the lab's services by the business units, and the opportunities that reached the market; see how to measure an innovation lab

A team from level three, a lab from level four. The Innovation Maturity Index in Innovation Mode 2.0 has seven levels. At level one, innovation-inactive, “there are no innovation-specific roles, teams, or capabilities in place”; at level two, innovation-aware, the leadership talks about it but there are only “occasional isolated events or small-scale, informal initiatives.” A lab at those levels is a showroom with no community to serve. At level three, innovation-engaged, people join workshops and “A small community of innovators is being formed,” but “most innovation efforts are ad hoc and not part of a managed innovation function.” That is the moment a function pays: a team and the first services. The place to test comes one level later, when workshops and prototyping are already happening.

• Level one: the work is awareness. The book's program opens with a Power-Up phase that “raises awareness about the importance of innovation, its methods, and tools.” Level two: enablers, knowledge sharing and a community, the Spark phase. A lab is on neither list
• Level three: the community exists and the method does not. Here the innovation team and its first services, manual if need be, turn ad hoc energy into a pipeline
• Level four, innovation-active: innovators “join hackathons, workshops, and prototyping activities,” and the company has innovation leadership “at the executive level, but not in the C-suite.” A lab, as a place to prototype and test, earns its keep here
• Level five and beyond: “A dedicated executive orchestrates the innovation function,” and the question is no longer whether to have a lab but how the labs, R&D groups and centers connect; see several labs
• Find your level: the Innovation Maturity Index, and how to use it to plan the next step

A team, with services the whole company can use. Innovation Mode 2.0 is direct about size: “Smaller companies may be at higher levels in the Innovation Maturity Index, even if they are using basic technologies and standard productivity tools. What matters most is the purpose, the culture, the talent, and the leadership's innovation attitude.” A mid-size company does not need a building to innovate; it needs a director, an innovation architect and a program manager, an agenda of problems worth solving, and a way to capture, assess and validate ideas. The book's own advice is to start manual: “investing heavily in technology is not a prerequisite for kick-starting the implementation of this framework.”

• The first three hires: the program director, an innovation architect and a program manager; the design guide has the roles and the order
• Services, not rooms: “certain services described in this framework could be offered by small, flexible teams or by leveraging the power of artificial intelligence directly.” Idea capture, assessment and a validation sprint fit in a shared document and a calendar
• Where a mid-size company wins: the distance between the lab and the business is short. The CEO can sponsor in person and the product teams sit down the corridor
• When a room helps: for workshops and prototyping once there is a community; a design sprint needs a wall, not a wing
• What scales with size: “the larger the organization, the more advanced the technology needed to streamline and effectively orchestrate the corporate innovation function.” The team comes first, the platform follows adoption

A startup is one; a scale-up needs discovery, not a lab. A startup is an innovation lab with a payroll; the whole company is the experiment. The question becomes real at scale-up, when the core product consumes everyone and the next opportunities stop being explored. Even there the answer is not a lab but a discovery capability: a small team that frames problems, assesses ideas against one model and runs validation, so the company keeps a portfolio of options while the main product grows. Innovation Mode 2.0 notes that the stage for radical innovation “is available to all, enabling relatively small companies, startups, or even individuals to disrupt the status quo”; the small company's advantage is that it needs no separate place to do it.

• Startup: the founders own discovery; the method matters more than the structure; see how to build an MVP with AI for the discipline that replaces the lab
• Scale-up: name one person who owns the opportunity portfolio beyond the core product, and give them a validation budget. That is the function in its smallest form
• The trap: an “innovation lab” in a scale-up usually means the founders' side projects with a new name. Run them through the same assessment as everything else
• What AI changed: a two-person team can now scout markets, synthesize research and prototype in days; the constraint is judgment, not capacity
• When the answer flips: after a few acquisitions, when the company becomes a group of businesses with their own agendas; then read the case for a network

A team, a network or a venture unit, and in some cases nothing yet. Innovation Mode 2.0 describes three models for organizing corporate innovation. The centralized model, where “This central team takes the form of technology labs or R&D departments,” suits science-heavy industries and risks a company that feels like “consumers” of the lab's ideas. The decentralized model lets innovation “originate from any team.” The federated model, the one the book builds on, runs “a network of ‘innovation cores’ established across the organization, which operate independently while orchestrated by an overall innovation program.” Each alternative is a different answer to the same question: who enables, who builds, and who connects.

• An innovation team with services: the core team of the book, whose mission is to enable; the smallest alternative and the one most companies actually need; see the innovation dream team
• A federated network: innovation cores in the divisions, one language, one central team for governance and method; see the federated model
• Embedded innovators without a center: cheap and popular, with “inconsistent innovation methods and gaps in knowledge sharing and cross-team collaboration” as the predictable cost; see the risks of embedding
• An innovation council: shared responsibility, which the book finds brings “diffused and ultimately diluted ownership and accountability”; see whether councils work
• A venture unit or venture client program: when the goal is new businesses rather than better ones; see when an internal venture unit works and the venture client model
• An AI center of excellence: a technology program, not an innovation function, and a poor substitute for one; see why the lab should not be an AI lab

No. R&D develops technologies; an innovation lab, in the sense of this guide, discovers and validates opportunities, which may or may not need new technology. In Innovation Mode 2.0, R&D is one node of the innovation network, alongside the product teams, the market-facing teams and the innovation team: in the book's scenario, the Portal notifies the network about “the release of a new technology update from the R&D team,” and the teams in the network decide what to do with it. Merging the two makes a research department with a demo wall; keeping them connected gives R&D a market and the lab a technology pipeline.

• Different questions: R&D asks what is possible; the innovation function asks what is worth doing and whether anyone would pay for it
• Different clocks: research runs on years and patents; opportunity validation runs on weeks and experiments; see how business experiments work
• Where they meet: the innovation agenda, which names problems worth solving, and the hub, where an R&D update becomes an input to the pipeline
• The centralized trap: a lab run as “technology labs or R&D departments” inherits R&D's distance from the business; the book's warning is that the rest of the company then feels like “consumers” of its output
• Science-heavy industries: keep both, connect them; the book accepts that “specialized laboratories or research facilities may coexist as part of a bigger innovation program”

Yes, if they run one system. A group with divisions, regions or acquired companies will grow several innovation groups whether or not headquarters approves, and the risk is not their number but their isolation. Innovation Mode 2.0 answers with the hub pattern: labs, discovery teams and innovation centers keep their own agenda, calendar and content, and share through the Innovation Portal. “This setup aligns the isolated groups, allows them to speak the same ‘innovation language,’ and empowers them to partner just in time.” Several labs on one method are a network; several labs on their own methods are a collection of silos with the same name.

• One language: the same idea and problem framing templates, the same assessment model, the same metrics, so opportunities can move between units and be compared
• Local autonomy: “each of the teams shown in Fig. 4.5 has full control over its own innovation agenda, calendar, and content,” and all contribute to the Global Innovation Agenda
• The central team: the federated model “requires a central innovation team that provides governance, resources, methodologies, and strategic direction”; without it, the network drifts apart
• Signals across the network: an idea from Sales and Customer Success, a technology update from R&D, a product update from the product teams, routed to the groups that can act on them
• Cost: a federated setup “requires strong leadership, well-defined decision-making frameworks, and powerful technology”; this is where the platform, not the building, earns its budget; see how the lab connects to the business units

Attention, and sometimes the rhythm of the business. The budget is the visible cost. The invisible ones are the time of people in the business units who join workshops, rotations and pilots; the management attention a visible lab draws; and the resentment a lab creates when it is seen as a privilege. Innovation Mode 2.0 names the fear precisely: executives “worry that innovation will become the exclusive domain of a specialized team rather than an organization-wide capability,” business units worry about “being underrepresented in the innovation process,” and employees “may believe they will be less empowered to innovate.” As I wrote in my post on the innovation team, innovation can disrupt the company before it disrupts any market, and the team's job includes preventing that.

• People's time: every workshop takes a day from a unit; budget it, and win the middle managers who release the people; see how to win middle managers over
• The noise: a lab that launches pilots into the business without coordination exposes the company's standard KPIs to risk; the mandate must include protecting the rhythm of operations
• The privilege problem: a team seen as enjoying “safe conditions without the pressure of production and operations-related tasks” loses the rest of the company; see what blocks a healthy innovation culture
• Leadership attention: a lab is the most visible thing an innovation program owns, so it absorbs the questions, including the ones about return on investment, which the book warns are too often answered with “inadequate or incomplete metrics”
• The cost of not having one: ideas with nowhere to go, problems nobody frames, and a company that learns about its market from its competitors' launches

Engagement within a year, outcomes later, and the order matters. Innovation Mode 2.0 gives the first phase of a well-sourced program three to six months to reach awareness and readiness, and notes that “A phase may last several months or even years.” The record agrees: Walmart's incubator ran nearly seven years before its capabilities were folded into the business; Walmart Canada's was integrated two years after it started, within a year of its official launch. An OECD OPSI analysis of public labs puts the time for innovations to scale at ten to twelve years. A lab judged on outcomes in year one is judged on nothing; one judged on activity in year three is theater.

• Year one: a community that returns, services the units ask for, a funnel with problems in validation; see what the first year should look like
• Late in year one, then year two: the first opportunities handed to product teams, the first experiments with customers reported in business terms
• Year three and after: outcomes: products, savings, ventures. “Many organizations evaluate long-term innovation using short-term KPIs,” and this is where it costs them
• Agree the clock in advance: the Innovation Masterplan fixes success criteria per phase; the book moves a program to the next phase “when a set of success criteria is consistently met,” not on a date
• The public-sector benchmark: “For innovations to scale, it is reasonable to calculate a time period of ten to twelve years.” Corporate clocks are shorter; the lesson about patience is the same

When the business stops asking, or when it graduates. There are two kinds of ending. In the first, no business unit has asked the lab for a service in a year, its output is demos, and its budget is defended with activity; close it, or rather convert it into the function it should have been: a small team with services. In the second, the lab's methods and people have been absorbed by the business, which is what Walmart described when it closed its incubator and what Nordstrom meant when it said innovation was “now everyone's job.” That is not failure. An OECD OPSI analysis of 137 public labs lists “closing a lab means failure” among the fallacies of the field.

• The warning signs: no unit has sent in a problem; no opportunity has crossed to a product team; the metrics are counts of ideas, workshops and patents, “vanity metrics of innovation activity” in the book's words
• Convert, do not just close: keep the agenda, the assessment model and the two or three people who ran the services; drop the space and the showcase. The function costs a fraction of the lab
• Graduation: Walmart's 300 people moved into its technology and innovation operations; Nordstrom moved parts of its lab into tech and business teams while keeping a core on specific customer opportunities. Plan for this ending from day one
• What to keep from a closed lab: the knowledge: the problems framed, the experiments run, the reasons ideas were dropped. The book's innovation function “produces, accumulates, and diffuses knowledge and innovation assets throughout the organization”; a closure that loses them is the real failure
• Who decides: whoever sponsored it, against the success criteria written into the Masterplan, not a new CFO reading a cost line; see how you know whether an innovation leader is succeeding

Less as a place, more as a function. In Innovation Mode 2.0 I put it this way: “When AI enables anyone to generate brilliant ideas, shape strategies, and soon build digital products, the ability to conceive and implement becomes less important.” Every product team can now prototype in days, so a lab whose value was the prototype has lost its monopoly. What becomes scarce is “separating signal from noise, recognizing genuine opportunities, racing to market,” and that is a function: an agenda, one assessment model for every idea, a portfolio and a path to market. The company needs more of that, in fewer rooms.

• What AI took from the lab: the prototype, the research summary, the workshop pre-read, the first draft of everything; the design guide lists what changes when AI does the work
• What AI gave the function: the book's “most vital innovation capability is a scalable opportunity discovery, validation, and realization function,” a system of cross-disciplinary teams “powered by AI that can pursue multiple opportunities simultaneously”
• The new risk: a hundred AI pilots in a hundred units, none framed as a problem, none assessed against the others. The function exists to stop that; see how to build a balanced AI portfolio
• Not an AI lab: the technology belongs inside the function, not on its door; see why the lab should not be an AI lab
• The decision, restated: if the question is whether we need a room where people build things, AI says less than before. If it is whether we need a system that decides what is worth building, AI says more

Five questions, then the smallest version that answers them. Is innovation on the leadership agenda with a sponsor who will show up? Where is the company on the Innovation Maturity Index? Does a validated opportunity have a team to go to? How many divisions innovate on their own? Do we want the lab to enable the company or to innovate for it? Write the answers into an Innovation Masterplan, the book's planning document, and let them size the first version: nothing yet, a team, a lab, or a network. Steve Blank's warning about innovation outposts applies to all of them: “Just establishing an Innovation Outpost doesn't mean that the corporation is innovating. At first it just means there's a new building.”

• No sponsor, or level one or two: nothing yet. Spend the year on awareness and a pulse survey; see how to build a culture of innovation
• Sponsor, level three, one business: a team of three with manual services and an agenda of problems; see how to start
• Level four, a community, a flow of problems, product teams that will take opportunities: a lab, designed as a function with a room, not a room with a function; see how to set up an innovation lab in the era of AI
• Several divisions already innovating: a network with a central team and one method; see several labs
• New businesses as the goal: a venture unit beside the function; see how to set up corporate venture building
• Whatever you choose: agree the clock and the exits in writing; see how long to give it and when to close or convert it