Issue#8: Zombie Pilots
The idea, the venture, the initiative that cannot die. Not because it is succeeding, but because stopping it is structurally harder than continuing.
It is on the portfolio slide again. No one knows why.
Most corporate innovation portfolios contain at least one idea, one venture, one initiative that everyone privately acknowledges is not working, but no one stops it.
Sometimes it is the quiet kind. Modestly funded, never celebrated, simply persisting. Quarter after quarter, it appears on the slide, absorbs a small allocation, and generates just enough activity to avoid the question no one wants to answer.
But there is a second variant that is harder to name and more expensive to ignore. This one is not starved of resources. It has dedicated headcount, meaningful budget, and sometimes even executive sponsorship. The team is busy. The updates are polished. And yet nothing has structurally moved. No market validation has landed. No revenue has materialised. No one has narrowed the scope or made a commitment that would actually put the initiative at risk of being tested by reality.
The resources keep flowing, not because the evidence justifies them. They continue because the commitment was made publicly. Now the sponsor’s credibility is attached to its continuation. The larger the investment, the harder it becomes to determine whether it is still warranted. This version of the zombie pilot does not appear to be neglected. It looks like conviction. That is what makes it dangerous.
Both are zombie pilots.
One starves quietly in the margin while the other feeds comfortably at the centre. Neither is alive in any meaningful operational sense. Neither is dead because no one has the authority, incentive, or structural mechanism to kill it. Together, they represent the most common and least discussed failure mode in corporate innovation governance.
The interesting question is not why these pilots fail. It is why they do not stop.
This is not a small problem. IDC’s research with Lenovo found that for every 33 proofs of concept launched across enterprise environments, only 4 reached production. The rest were not cancelled. They had not progressed. They existed in what the researchers described as a permanently provisional state, consuming budget and eroding executive confidence without ever reaching a clear resolution. That description is precise. It is also a description of a governance failure, not a technology failure.
Built to start. Not built to stop.
The obvious explanation is sunk cost. Someone invested political capital to launch the pilot, and stopping it publicly means writing off that investment. But sunk cost is a psychological explanation. It is incomplete. The deeper problem is structural.
Most innovation governance systems are designed to do one thing well … approve. It might not always be as efficient as operators would like, but of there are clear stage gates for funding. They have review boards for launch. They have sponsors who champion initiatives through internal resistance. The entire architecture is oriented around getting things started.
In Issue #7, I examined why well-resourced innovation functions still fail to produce ventures. Systems designed around activity metrics rather than execution outcomes create what amounts to “innovation theatre”. The zombie pilot is a specific expression of that same structural condition. The function can start things. It can run things. What it cannot do is stop things, because the stop authority was never built into the governance architecture.
Stop authority, the clearly assigned right to kill an initiative, is the most absent governance capability in corporate innovation. Not because leaders lack courage, but because the system was never designed to require it. There is no stage gate for cessation. No review trigger that asks “should this still exist?” No role whose explicit accountability includes recommending termination.
Multiple research programmes have identified the same gap from different angles. What researchers at RT Insights, Bain Capital Ventures, and others independently describe as an “ownership vacuum,” where the experiment has a sponsor but the outcome does not have an owner, is the structural condition that allows zombie pilots to persist. Without an owner accountable for the outcome, there is no one with the standing to recommend cessation. The vacuum is not an oversight. It is a predictable result of governance systems that were designed to launch initiatives, not to resolve them.
Without a stop authority, the default state of any initiative is continuation.
This matters because continuing is not neutral. Every zombie pilot, whether it is starving in the margin or feeding at the centre, consumes more than its line-item budget. It occupies a portfolio slot that signals “we are already doing something in that space.” It absorbs management attention, even if only the attention required to avoid deciding it. And it provides cover for the sponsor who launched it, the executive who approved it, and the team that would otherwise need redeployment.
The well-funded variant compounds this. The scale of investment creates its own justification, and the more the organisation has committed, the harder it becomes to ask whether the commitment is still warranted. At some point, the sunk cost becomes the strategy.
The zombie pilot does not survive because it is succeeding. It survives because the cost of stopping it is borne by an individual, while the cost of continuing the venture is distributed across the organisation. Stopping requires someone to raise their hand, absorb the political risk, and trigger a process that probably does not exist.
Continuing requires nothing. It is the organisational equivalent of leaving a subscription running, because cancelling it means finding the login details.
Courage is not the fix.
The implication is not “be braver about killing pilots.” That is a mindset recommendation disguised as strategy, and it misses the point entirely. If stopping required only courage, organisations with strong leaders would never have zombie pilots. They do.
The implication is that stop authority must be designed into the governance system with the same rigour as start authority. It needs to be structural, not heroic.
This means several things in practice.
Every pilot approved should have a pre-committed termination trigger.
Not a vague milestone review, but a specific condition under which the default shifts from persistence to cessation. The question at review should not be “is there a reason to stop this?” It should be “has this earned the right to continue?” The burden of proof matters enormously. When continuation remains the default, evidence of failure must accumulate to an extraordinary threshold before anyone acts. When cessation is the default, even modest initiatives must periodically demonstrate that they deserve the resources they consume.The stop authority must be assigned to a role rather than left to emerge.
Someone in the governance structure needs the explicit right, and the explicit expectation, to recommend kills. Not as a punishment. Not as a failure of sponsorship. As a routine function of portfolio discipline. The absence of this role is why zombie pilots cluster. No one’s job includes the line item “identify initiatives that should no longer exist.”The cost of continuation must be made visible.
Zombie pilots persist partly because their cost is illegible. The budget line looks modest. But the portfolio slot they occupy, the strategic space they block, the opportunity cost of the team’s time, the signal they send about what the organisation tolerates. However, none of this shows on a dashboard. Making the full cost of continuation visible is a governance design choice. It does not happen by accident, and its absence is not neutral.
The deeper pattern here extends well beyond innovation portfolios.
Organisations are generally better at creating than at ceasing. They are better at launching than at landing. The muscles for initiation are well-developed. The muscles for termination are atrophied, not because of disuse, but because they were never built.
Having a stop authority is not an optional governance refinement. It is a structural prerequisite for portfolio discipline. Without it, every portfolio will accumulate drag. Not because the people are wrong, but because the system makes stopping harder than continuing.
And in any system where stopping is harder than continuing, things that should stop will not.
The Stop Authority Diagnostic
The pattern described above can be diagnosed. Pilots that persist not because they are succeeding but because stopping them is structurally harder than continuing. It does not require a transformation programme or a governance overhaul to identify. It requires five questions.
These questions are not designed to evaluate whether a specific pilot is good or bad. Rather, they are designed to assess whether your governance system has the structural capacity to stop anything.
Is a stop authority assigned?
For any active initiative or venture in the portfolio, can you name the person whose role includes the explicit authority to recommend termination?
Not the sponsor.
Not the executive who approved it.
A role with the standing to say “this should no longer exist” without that statement being read as a political act. If you cannot name that person, the stop authority is absent. It does not matter how strong your leaders are. The capability has not been built.
Where does the burden of proof sit?
When an initiative comes up for review, what is the default?
Does the team need to demonstrate that the pilot has earned continuation, or does someone need to build a case for stopping it?
If the burden of proof sits with cessation, if you need a reason to stop but not a reason to continue, the system is architecturally biased toward accumulation.
Zombie pilots are the predictable result.
Do termination triggers exist before launch?
When the pilot was approved, was there a pre-committed condition under which it would end?
Not a milestone.
Not a KPI target.
A specific observable condition: “if X has not occurred by Y, the default is termination.” If the answer is no, the pilot was launched with an open-ended claim on resources.
Every venture pursued without a termination trigger is a future zombie candidate.
Is the full cost of continuation visible?
Can you see, on a dashboard, in a review pack, anywhere, the total cost of an active pilot beyond its budget line?
The portfolio slot it occupies. The strategic space it blocks. The management attention it absorbs. The opportunity cost of the team assigned to it. If only the direct budget is visible, the cost of continuation remains illegible, and illegible costs cannot drive decisions.
When was the last kill?
Not a pivot. Not a “strategic redirection.” Not a quiet de-funding that let something expire without anyone acknowledging what happened.
An explicit, deliberate decision to stop an initiative and communicate that decision as a governance outcome. If you cannot point to one in the past twelve months, the stop muscle has not been exercised. It may not exist.
Understanding the pattern
These five questions do not produce a score, but rather they produce a structural picture.
If you answered “no” to three or more, your governance system lacks functional stop authority. You have a system that can start things, but cannot stop them. Your portfolio will accumulate drag, not because of bad judgement, but because the architecture makes continuation the path of least resistance.
Each gap the diagnostic reveals maps to a specific governance design choice: assigning the authority, shifting the burden, pre-committing the trigger, or surfacing the cost. These are not cultural goals. They are built decisions.
The fix is not about becoming more “ruthless” or “disciplined” in the abstract. It is about building the specific governance capabilities that make stopping as operationally routine as starting.
The muscles that were never built
Zombie pilots are a symptom of a bigger issue. Organisations have spent decades improving the architecture of initiation with stage gates, investment committees, innovation sprints, accelerator programmes and even venture boards. The infrastructure for starting things is sophisticated, well-resourced, and culturally celebrated. However, the infrastructure for stopping things barely exists.
This asymmetry is not limited to innovation portfolios. It appears in product lines that no longer serve the strategy. In committees that outlived their purpose. In reporting structures that persist because no one has the standing to dissolve them. The inability to cease is an organisational condition, and innovation portfolios are where it becomes most visible, because the gap between activity and value is hardest to ignore.
Scott Kirsner observed in Harvard Business Review that when a CEO reveals a major innovation initiative, you should mark your calendar. Three years later, many of these ventures will have quietly expired without an obituary. That quiet expiry is the zombie pilot’s final form. Not a dramatic failure. Not an explicit decision. Just a slow fade that no one is accountable for, because the system never assigned accountability for cessation in the first place.
I have written separately about how this pattern is playing out across Australia’s government, enterprise, and venture portfolios, where audit bodies and digital transformation agencies have documented the same structural failure for years.
The language varies, including “Pilotitis” or “Pilot purgatory.” The underlying condition is identical, that systems that can start things but cannot force a scale-or-stop decision once a pilot has done the job it was meant to do.
Execution discipline is not only about doing things well. It is about stopping things cleanly.
When stopping an initiative is structurally harder than persisting with it, the portfolio does not have a performance problem. It has a governance problem.
References & Further Reading
IDC, research undertaken in partnership with Lenovo, CIO Playbook 2025: 88% of AI proofs of concept fail to reach production. https://www.cio.com/article/3850763/88-of-ai-pilots-fail-to-reach-production-but-thats-not-all-on-it.html
Scott Kirsner, The Stage Where Most Innovation Projects Fail, Harvard Business Review, 2017. https://hbr.org/2017/04/the-stage-where-most-innovation-projects-fail
SoftwareSeni, The Enterprise AI Pilot Purgatory Problem: What the Statistics Actually Tell Us, 2026. Synthesises research from IDC, MIT NANDA, McKinsey, BCG, RT Insights, Bain Capital Ventures, and others on the “ownership vacuum” pattern. https://www.softwareseni.com/the-enterprise-ai-pilot-purgatory-problem-what-the-statistics-actually-tell-us/
The Venture Operator, Observation: Australia’s Zombie Pilot Problem, Put Ideas To Work, 2026. https://www.putideastowork.com/p/observation-australias-zombie-pilot
The Venture Operator, Issue 07: When Nothing Changes, Put Ideas To Work, 2026. https://www.putideastowork.com/p/when-nothing-changes
This article is based on the author's professional experience and interpretation of publicly available information. It is provided for general informational purposes only and does not constitute advice. Any views expressed are the author's own and do not refer to any specific organisation, program, or individual.
This is Issue 08 of Put Ideas To Work. It builds on the structural patterns explored in Issue 07: When Nothing Changes, which examined why well-resourced innovation functions still fail to produce ventures. If the Stop Authority Diagnostic is useful, share it with someone who runs a portfolio review. It takes five minutes and the pattern will be obvious. And if you are not yet subscribed, Put Ideas To Work publishes biweekly. Each issue builds a diagnostic framework for venture execution under constraint.