Most corporate ventures do not die in the market. They die in meetings. A venture that had early customers and a credible team loses its budget in the annual plan, has its engineers recalled to fix a problem in the core business, waits four months for a procurement approval, or is asked to show a return on capital in its second year. By the time anyone asks whether the product was any good, the venture has already been starved.
That pattern is a failure of corporate innovation governance, the set of rules that decide who can fund, staff, approve, stop and measure a new venture inside an established company. Governance sounds administrative, but it is the main reason the same idea can succeed as a startup and fail as a corporate project. A startup is governed by a small board and a bank balance. A corporate venture is governed by the processes of a business designed for something else entirely.
This article explains why those processes are hostile to new ventures, using the research that has studied the problem most carefully, and then sets out a governance design checklist covering decision rights, funding cadence, kill rules, talent, asset access and measurement. It is written for CEOs, boards, chief strategy and innovation officers and the leaders of corporate ventures. The choice of studio model and the comparison between venture building and other instruments are covered in separate PanScience Research pieces in this cluster; this piece focuses on what happens once a venture exists.
Why the parent's machinery works against the venture
Vijay Govindarajan and Chris Trimble, of Dartmouth's Tuck School, describe the core of an established company as a "performance engine": an organisation built for efficiency, accountability and predictable results. In the summary of their book The Other Side of Innovation published by Tuck, they put the problem bluntly: "Organizations are not designed for innovation, they are designed for ongoing operations," and "innovation and ongoing operations are always and inevitably in conflict." They do not argue for dismantling the performance engine, which they call "incredibly powerful." They argue that a new venture needs a different structure alongside it, and that most companies get that structure wrong.
The most common error, in their account, is that new units copy the parent's routines and become "little performance engines," which they describe as "deadly to innovation." A venture that inherits the parent's planning cycle, approval thresholds, hiring grades and performance metrics will behave like a small, slow division, whatever its mandate says.
Charles O'Reilly and Michael Tushman's April 2004 Harvard Business Review article, "The Ambidextrous Organization," remains the most cited evidence on which structures work. They studied 35 attempts to launch breakthrough innovations undertaken by 15 business units in nine industries, and classified them by organisational design. Seven were run within existing functional designs, nine as cross-functional teams, four as unsupported teams and fifteen within what they called ambidextrous organisations. The results were stark. More than 90% of those using the ambidextrous structure succeeded in their attempts. Only 25% of those using functional designs reached their goals. None of the cross-functional or unsupported teams did.
The ambidextrous design had two features. The new, exploratory units were separated from the traditional, exploitative ones, so each could have its own processes, structures and culture. And the separation was managed through tight links at the senior executive level: in the authors' words, "a tightly integrated senior team." The sample is small and the study is two decades old, so the percentages should not be treated as general success rates, but the mechanism it identifies is clear: separation without senior integration leaves the venture orphaned, and integration without separation leaves it smothered.
The evidence from corporate-startup work
More recent survey evidence points in the same direction. McKinsey's October 2024 survey of 1,176 executives on corporate venture building found that experienced builders were 2.6 times more likely than novices to have dedicated funding, 2.3 times more likely to have a dedicated leader, 1.4 times more likely to have C-suite champions and three times more likely to have a formal venture-building process. Each of those is a governance choice, not a product choice.
McKinsey's May 2021 article on collaborations between corporates and startups reported that only 28% of startups were satisfied with their corporate partnerships and that the majority of CVC units stop investing after two to three years. Among the reasons it gave for collaborations derailing were weak internal sponsorship, unclear strategic goals, slow corporate processes, a lack of impact tracking and getting stuck in pilots with no route to scale. Again, these are governance failures.
Seven ways a venture dies inside the parent
Drawing on the research above and on common practice, the failure modes can be grouped into seven. This classification is the author's synthesis.
Starvation by budget cycle. The venture is funded annually alongside the core, and in a bad year it is the easiest line to cut because it has the least revenue.
Recall of talent. The venture's best people are borrowed from the core and are recalled when the core needs them, or they return voluntarily because their career path runs through the core.
Process drag. The venture must follow the parent's procurement, legal, IT security and hiring processes, which are designed for scale and take months.
Wrong metrics. The venture is measured on revenue, margin or return on capital too early, before those numbers can mean anything.
Asset denial. The business units that own customers, data or plants decline to give the venture access, because it creates work or risk for them and no reward.
Sponsor loss. The executive who championed the venture moves on, and nobody else owns it.
Undead ventures. The opposite failure: nobody is empowered to stop a venture that is not working, so it consumes budget and talent for years.
Each of these has a governance remedy, and the checklist below is organised around them.
A corporate innovation governance checklist
The table is the article's main analytical contribution. For each dimension it sets out the design choice, a test question a board can ask, and a red flag that suggests the design is failing. It assumes a corporate running one or more ventures that are separate from the core business but still majority-owned.
Applying the checklist
Three principles help in using the checklist.
Separate the routine, integrate the leadership. This is the O'Reilly and Tushman finding restated as a design rule. The venture should have its own processes for hiring, procurement, metrics and planning, and its leadership should report into a senior executive who also has authority over the business units that own the assets it needs. Govindarajan and Trimble make a similar point in different language: the dedicated team should be built "as if creating a new company," but it must work in partnership with the shared staff of the performance engine, and senior executives must actively make that partnership work.
Decide the kill rule before the stage, not during it. Kill decisions taken during a stage are distorted by sunk cost and by the sponsor's reputation. A kill rule set at the start of a stage, with named evidence and a named decision-maker, turns stopping a venture into a normal outcome rather than a failure. It also protects the venture: a venture that meets its stage criteria has a stronger claim on the next tranche than one that is merely liked.
Pay the business units. Asset denial is often treated as a cultural problem, but it is usually an incentive problem. A business unit head who shares customer access with a venture takes on risk and work and gains nothing. Governance should give business units a share in the venture's success, through internal transfer pricing, a share of revenue or recognition in their own targets.
How the checklist fits Indian corporates
Three features of Indian corporates affect how the checklist is applied.
In promoter-led groups, decision rights are often concentrated in the promoter family or a small group. That can make senior integration easier, because one person can resolve conflicts quickly, but it also makes sponsor continuity fragile and can blur the line between the venture board and the group's informal decision-making. A written venture charter matters more, not less, in that setting.
In listed companies, venture-linked incentives face constraints from employee stock option rules and from the difficulty of valuing a subsidiary. Phantom equity and equity in a separate venture entity are the common workarounds; incentive design for intrapreneurs is covered in a separate PanScience Research piece.
In companies where IT security and data protection approvals are centralised, process exemptions need care. A venture cannot be exempted from the Digital Personal Data Protection Act or sector rules, but it can be given a pre-approved architecture and a fast-track review that applies the same standards with less delay.
A worked illustration
Consider how the checklist applies to a venture built around an enterprise AI product for a regulated industry. Eigen, a venture in the PanScience portfolio, describes itself as building on-premise AI for regulated industries. A corporate building a comparable venture internally would meet several of the seven failure modes at once: the venture would need data from business units that are wary of sharing it (asset denial), security approvals that run on the core's timetable (process drag), and engineers who could be pulled back into the core's own IT projects (recall of talent). The checklist turns those predictable problems into design decisions taken before launch: a signed data access agreement, a pre-approved deployment architecture and a dedicated team with its own reporting line. None of this guarantees the product will succeed in the market. It ensures that, if it fails, it fails in the market.
Common objections
"Separation creates a two-tier company." It can, if the venture is seen as privileged. The remedy is to make the venture's terms explicit and its stage gates visible, and to share its success with the business units that support it.
"We cannot exempt anyone from our processes." Exemption does not mean lower standards. It means a different route, often with the same reviewers, sized to the venture's risk. A pilot contract worth a small sum should not need the same approval path as a multi-year enterprise deal.
"Kill rules will make people timid." The opposite is more likely, on our reading. When stopping is a normal, pre-agreed outcome, teams take bolder bets because failure does not end careers. The career-guarantee row of the checklist exists for exactly this reason.
Conclusion
Corporate ventures fail inside the parent for reasons that are predictable, well documented and largely within the parent's control. The research is consistent: ventures need separation from the core's routines and integration at the top; they need dedicated funding, dedicated leaders and formal processes; and they need business units that are rewarded, not merely instructed, to share their assets. Corporate innovation governance is the discipline of writing those requirements down before the venture starts, so that the venture is judged on whether customers want what it builds rather than on whether it survived the annual plan. A board that works through the checklist above before approving its next venture will not guarantee a success, but it will remove the most common reasons for failure that have nothing to do with the market.
Methodology and sources
The ambidexterity study is Charles A. O'Reilly III and Michael L. Tushman, "The Ambidextrous Organization," Harvard Business Review, April 2004 (https://hbr.org/2004/04/the-ambidextrous-organization); study design and results were read from a full-text copy hosted for teaching by the University of Trieste (https://moodle2.units.it/pluginfile.php/577953/mod_resource/content/1/Oreally_Tushman-2004- The%20Ambidextrous%20Organization.pdf). The sample is 35 initiatives and the findings are the authors' classification. Govindarajan and Trimble's quotations are from the Tuck School's summary of The Other Side of Innovation (https://mba.tuck.dartmouth.edu/pages/faculty/chris.trimble/osi/press/faqs.html). Survey figures are from McKinsey, "How CEOs are turning corporate venture building into outsize growth," 22 October 2024 (https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/how- ceos-are-turning-corporate-venture-building-into-outsize-growth) and McKinsey, "Collaborations between corporates and start-ups," 10 May 2021 (https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/collaborations-between-corporates-and-start-ups); both are self-reported survey results. The seven failure modes, the governance checklist and the worked illustration are the author's analysis and are illustrative, not observed data. The Eigen description is company-reported (https://panscience.xyz).