The Pilot Worked. Who Owns It on Monday?

• A validated pilot is not a decision
• Not resistance, but two operating logics
• What the pilot did not establish
• Fragmentation and structural gates
• What the missing handover produces
• The Nordic dimension
• The Structural Response
• Conclusion
• References

A pilot proves that a solution can work; it does not establish who will run it, who will pay for it, or when the decision to continue gets made.

A validated pilot is not a decision

Most large organisations have learned to run pilots. Very few have made the step that follows a pilot routine. McKinsey surveyed more than a thousand companies on operating-model change and found that although most had made progress in modernising operations, fewer than a third had moved beyond the pilot phase, and that the position had barely shifted over the preceding two years. The term the firm coined for the resulting accumulation of stalled experiments, pilot purgatory, has been in use since 2018. That it remains current is the point worth noticing: this is not a new problem awaiting a solution, but a stable structural condition that most organisations have not yet designed against.

Bridgium’s second research report, From Discovery to Practice, examines where corporate and startup collaboration gains and loses momentum. It draws on 48 qualitative interviews conducted between September and December 2025: 28 with innovation leaders and senior practitioners in large Nordic and European corporations, and 20 with founders and senior representatives of startups and SMEs already working with them. The report treats collaboration as a staged Innovation Flow with three stages, and the third of them, Integration and Embedding, is where the interviews are most consistent. One respondent stated the finding in a single sentence:

“Even successful pilots don’t automatically have a place in the organization.”
— Strategy Director · Telecommunications · Denmark

Read slowly, that sentence separates two things that innovation portfolios routinely treat as one. Technical validation establishes that a solution performs. Organisational legitimacy establishes that the organisation will carry it. The second does not follow from the first, and nothing in a well-run pilot produces it as a by-product. This article sets out what the research says has to be in place instead, drawn from section 7 of the report.

Not resistance, but two operating logics

The usual account of what happens after a successful pilot is a story about resistance: operations is conservative, the business unit protects its routines, middle management blocks change. The interviews do not describe resistance. They describe two groups applying different and equally defensible standards to the same result.

Innovation and transformation teams read a validated pilot as evidence for scaling and continuation. Operational units evaluate the same solution through continuity, process fit, resource allocation and long-term maintainability. Those criteria are not obstacles that someone invented; they are the reason the operation is reliable. A unit that adopts every validated solution offered to it will spend its year integrating rather than operating.

“What works in a pilot does not always fit into existing processes.”
— Operations Director · Logistics · Sweden

The incentive structure reinforces the split. Kerr’s account of rewarding one behaviour while hoping for another describes the position of an operational leader precisely: measured on uptime, cost per unit, service level and compliance, then asked to absorb the integration risk of an unproven solution alongside those measures, with the credit for the innovation attaching elsewhere. The behaviour that follows is rational. What the organisation is actually asking for is not more openness from the operations side but a change in what the operations side is accountable for, and that is a design decision rather than a matter of persuasion.

O’Reilly and Tushman’s work on ambidextrous organisations identifies the same structure and its condition of success: exploratory and exploitative units can be separated, but the separation only produces value where senior management actively integrates the two, including through targets and resource allocation. Where the innovation unit is separated and left to negotiate its own transfers, the separation produces exactly the pattern the interviews describe.

“Projects lose momentum when they move from innovation teams to business units.”
— Head of R&D · Energy · Norway

What the pilot did not establish

The report identifies the absence of clear embedding mechanisms as a recurring issue: even successful pilots often lack explicit ownership, operational resources and structured pathways toward implementation. Budget ownership is named specifically. A pilot can be technically validated and still fail to move if no business unit, P&L owner or operational sponsor is prepared to fund scaling, allocate people or absorb implementation risk.

The most useful diagnostic here is subtractive. Rather than asking what the pilot proved, ask what it left open.

Table 1. A subtractive reading of pilot results. Constructed from Bridgium, From Discovery to Practice (2026), sections 7.1 and 7.3.

Two respondents describe what fills the gap when these questions are not answered. The first names the mechanism:

“Information moves quite well, but decisions are still made within units. That’s where things slow down.”
— Digital Innovation Leader · Telecommunications · Finland

The second names the symptom. Asked what has to happen for a validated solution to move, a respondent framed it as two direct questions to the receiving organisation:

“Do you understand how this future could look? And are you willing to commit to it?”
— Director of Growth & Development · Energy Technology · Netherlands

Without that step, the same respondent observes, initiatives receive encouragement of the “this is nice, please carry on” kind while no one has a real intention of putting resources behind them. This is worth treating as a category of outcome rather than as a communication failure. Encouragement is cheap, it is sincere, and it is indistinguishable from commitment until the moment resources are requested, which is usually several months after the pilot has ended and the innovation team has moved on.

Fragmentation and structural gates

Embedding a solution requires coordination across innovation teams, operational units, procurement, legal, compliance and management structures. The report describes the resulting communication problem as internal fragmentation, and one respondent gives it its plainest form:

“Different teams evaluate the same startup differently, and there is no shared view.”
— Innovation Program Manager · Automotive · Germany

This is not a disagreement that better information solves. Weick’s work on sensemaking is useful here because it describes interpretation as something organisations do collectively and retrospectively rather than as a matter of receiving facts. Procurement, IT security, the operating unit and the innovation team are not looking at different data about the same solution; they are constructing different objects out of the same data, each valid within its own function. Burt’s account of structural holes explains why nothing reconciles them by default: where no one occupies the position between functions, no shared view forms, because forming one is somebody’s work and it has not been assigned.

Alongside fragmentation, the report identifies a second and more predictable obstacle. At the Integration stage, organisations prioritise operational stability, and procurement, legal, IP, cyber and data-security vetting become structural gates between pilot validation and operational adoption. For the external partner, this marks a shift from proving relevance to proving reliability, and it typically arrives without warning at the point where everyone believed the difficult part was over.

These gates are not discretionary and they are not fast. In financial services, the Digital Operational Resilience Act sets specific requirements for contractual arrangements with third-party ICT providers, which means a small supplier entering a production environment enters a documented regime rather than a purchase. The gates are entirely anticipable. What the interviews suggest is that they are rarely anticipated during the pilot, when anticipating them would cost almost nothing.

Table 2. The four Innovation Flow conditions at the Integration stage, expressed as questions. Conditions and requirements from Bridgium, From Discovery to Practice (2026), Table 5 and section 7.5; question framing by the Bridgium research team.

What the missing handover produces

Four outcomes recur in the interviews, each traceable to a specific thing that was not established while the pilot was running.

• Encouragement without allocation. Where no one is asked to commit resources at a named moment, support remains verbal. The initiative is praised, kept alive in reporting and never funded. Because nobody has refused anything, the situation produces no signal that would trigger a review.
• The gap between initiator and operator. Pilots are frequently initiated by innovation or transformation teams, while operational implementation depends on business units responsible for infrastructure, execution, budgets and long-term support. Where the receiving unit has not been identified during the pilot, validation produces a result with no destination, and the transfer becomes a negotiation between parties with different accountabilities and no shared target.
• Divergent evaluation without reconciliation. Multiple functions assess the same solution against different criteria, all of them legitimate. Without a forum and an owner for reconciliation, the assessments do not converge; they accumulate, and each one is sufficient to delay the decision on its own.
• Learning that stays local. At the Integration stage learning becomes distributed across teams, countries, operational contexts and user groups, and some units adapt quickly while others meet resistance or different constraints. Much of what is learned during a first implementation is tacit in Nonaka and Takeuchi’s sense, held in the practice of the people who did it rather than in any document. Where no mechanism converts it, the organisation restarts integration learning rather than accumulating the capability to embed anything at all. This is what the report means by innovation memory, and its absence is the most expensive of the four, because it is paid again at every site.

The Nordic dimension

The regional pattern here contains a genuine advantage and a specific risk, and they have the same origin.

The advantage is horizontal information flow. Flat structures, low power distance and a habit of direct cross-unit contact mean that in the Nordic organisations in the sample, knowing about a pilot in another unit is rarely the constraint. The respondent quoted earlier says exactly this before naming the problem: information moves quite well. That is not a small asset. It removes an entire class of obstacle that more hierarchical organisations spend considerable effort on, and it makes the connectivity condition easier to satisfy than it is elsewhere in Europe.

The risk sits directly behind it. In the same sentence the respondent notes that decisions are still made within units, and that is where things slow down. Where decision rights are distributed across units and coordination is consensual, the absence of a named owner is difficult to see, because the process does not stop. Meetings continue, everyone agrees the solution is good, and no unit has said no. Broad agreement is not the same thing as allocated responsibility, and an organisation that is good at the first can go a long time without noticing that it has not produced the second. In regulated and asset-heavy sectors, which describe much of Nordic industry, the delay compounds: procurement, security and compliance gates add months that were not planned for, and asset lifecycles measured in decades mean the next window for the change may be a long way off.

The practical implication is not that Nordic organisations should become more hierarchical. It is that consensus-based coordination has to be paired with an explicit assignment of ownership and a dated decision, because the coordination style will not generate either on its own.

The Structural Response

Four responses follow from the research for leaders running innovation portfolios in large organisations.

1. Name the receiving unit before the pilot starts. Not the sponsor and not the innovation team, but the unit that would operate the solution and carry its cost. If that unit cannot be named at the outset, the pilot is an experiment about the technology rather than a step toward adoption, and it should be scoped and budgeted as one. Both are legitimate. Confusing them is what produces the accumulation.
2. Make continuation a scheduled decision, not a budget outcome. A pilot should end on a date with a named decision-maker and three available answers: scale, extend, stop. Where the continuation decision is left to arrive through the annual planning cycle instead, the solution competes months later against priorities set in a different context, and the pilot result carries little weight in that comparison.
3. Bring the gates forward. Procurement, legal, IP, cyber and data-security review can begin while the pilot runs rather than after it succeeds. Running them in parallel costs a few hours of coordination during the pilot and removes several months from the transition, and it surfaces disqualifying constraints early enough that the pilot can be redesigned around them.
4. Treat the handover as a transfer of ownership, not a report. A final presentation transfers information. What has to move is accountability, along with the budget, the support arrangement, the feedback loop and the record of what was learned and what failed. Where implementation learning is captured in a form the next unit can use, the second adoption is materially cheaper than the first. Where it is not, the organisation pays the full price every time.

Conclusion

The finding in this section of the research is unusually specific, and that is what makes it useful. Collaboration does not typically break down because a solution was wrong or because a partner underperformed. It breaks down at a transition that no one owns, between a team whose accountability ends at validation and a unit whose accountability never formally began.

Everything required to prevent this is available during the pilot, when it costs almost nothing: naming the receiving unit, dating the decision, opening the gates in parallel, and writing down what was learned in a form someone else can use. Everything required to repair it afterwards is expensive, because by then the sponsor has moved on, the budget cycle has closed and the external partner has drawn its own conclusion about how the organisation works.

So the question is narrow and answerable today. For the pilot closest to completion in your portfolio, can you name the unit that will run the solution, the budget line that will fund it, and the date on which the continuation decision will be made?

Bridgium’s second research report, From Discovery to Practice: How Corporate–Startup Collaboration Becomes Usable, sets out the integration-stage findings in full, together with the practice cases behind them:
bridgium-research.eu/startup-report-2026

References

1. Bridgium Research Team (2026). From Discovery to Practice: How Corporate–Startup Collaboration Becomes Usable. Illarionova N., Verlin K., Verlin A. Report
2. Bridgium Research Team (2026). How Innovation Happens: Insights from Leading Enterprises in Times of Change. Illarionova N., Verlin K., Verlin A. Report
3. Kerr, S. (1975). On the Folly of Rewarding A, While Hoping for B. Academy of Management Journal, 18(4), 769–783. JSTOR
4. O’Reilly, C. A. and Tushman, M. L. (2004). The Ambidextrous Organization. Harvard Business Review, April 2004. Read
5. March, J. G. (1991). Exploration and Exploitation in Organizational Learning. Organization Science, 2(1), 71–87. JSTOR
6. Weick, K. E. (1995). Sensemaking in Organizations. Thousand Oaks, CA: Sage. Publisher
7. Burt, R. S. (1992). Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press. Publisher
8. Nonaka, I. and Takeuchi, H. (1995). The Knowledge-Creating Company. New York: Oxford University Press. Publisher
9. Berger, P. L. and Luckmann, T. (1966). The Social Construction of Reality. New York: Doubleday. Publisher
10. McKinsey & Company (2020). Breaching the great wall to scale. Read
11. McKinsey & Company. How digital manufacturing can escape pilot purgatory. Read
12. Regulation (EU) 2022/2554 on digital operational resilience for the financial sector (DORA). EUR-Lex