A corporate venture team puts $5m into a promising startup. Eighteen months later, a finance review asks the only question the company knows how to ask, "what's the return?", and the honest answer is "we don't know yet." The programme gets quietly defunded. The problem was never the bet. It was that the company used a ruler built for mature businesses to measure something that didn't have a single year of normal numbers to its name.

The quick version

  • Corporate venturing is how an established company pursues growth outside its core, and corporate venture capital (CVC) is the version where it takes minority equity in external startups.
  • The evidence is blunt: CVC reliably delivers strategic value (a window on new technology and markets), but on average shows no significant financial out-performance versus independent VC.
  • Innovation accounting, Eric Ries's idea, is how you measure a venture when revenue, profit and market share are all roughly zero: track leading indicators of value and growth, not forecasts.
  • The fix is unglamorous: decide whether each bet is strategic or financial before you fund it, then judge it on the metrics that match. Mixing the two is how good programmes die.

The idea in depth

Start with the vocabulary, because the two terms get used loosely and it matters. Corporate venturing is the umbrella: any structured way a large firm reaches for growth beyond its existing business, building internal ventures, running an accelerator, forming joint ventures, acquiring, or investing. Corporate venture capital is one branch of that tree: the company takes a minority stake in an outside startup, usually through a dedicated fund or unit, the way Google, Intel, Salesforce and dozens of others do.

This is not a niche activity. According to CB Insights' State of CVC 2024 report, global CVC-backed funding rose 20% year on year to $65.9bn across 3,434 deals, with AI startups taking a record 37% of CVC dollars. Corporates are pouring real money into other people's startups. The question is what they should expect back.

The cleanest map of "why invest" is still Henry Chesbrough's, from his 2002 Harvard Business Review article "Making Sense of Corporate Venture Capital." He sorts investments on two axes, the objective (strategic vs financial) and the operational link to the company's own business (tight vs loose), producing four modes: driving (strategic + tightly linked), enabling (strategic + loosely linked), emergent (financial today but strategically promising) and passive (financial, loosely linked, and, Chesbrough argues, the hardest to justify, because a corporate balance sheet has no edge as a pure financial investor). Label every prospective deal with one of those four before you write the cheque, because the label decides which metrics are allowed to judge it later.

quadrantChart
    title Why is the company really investing?
    x-axis "Loose link to core" --> "Tight link to core"
    y-axis "Financial objective" --> "Strategic objective"
    quadrant-1 "Driving"
    quadrant-2 "Enabling"
    quadrant-3 "Passive"
    quadrant-4 "Emergent"
Chesbrough's four modes of corporate venture investment, by objective and operational link. Leaders Loop

What the evidence actually says about returns

Here is the uncomfortable finding the brochures skip. A 2023 meta-analysis in The Journal of Technology Transfer by Patrick Haslanger, Erik Lehmann and Nikolaus Seitz pooled 32 CVC studies (over 105,000 observations) and reached a two-part conclusion: CVC investment is positively linked to strategic performance for both startups and investors, but shows no significant relationship with financial outcomes. In plain terms, corporate venturing tends to pay off as a learning and access machine, not as a money-printing one.

That squares with the foundational empirical work. Paul Gompers and Josh Lerner, examining tens of thousands of venture investments (NBER, 1998), found corporate-backed investments were at least as successful as independent-VC ones, but specifically where there was strategic overlap between parent and startup. Their sharpest point is about survival: programmes without a strong strategic focus were far less stable and frequently ceased after only a few deals, while strategically focused ones were as durable as independent funds. So stop benchmarking your CVC unit against a financial VC's IRR, and start asking what it taught the core business, partnerships formed, technologies de-risked, markets entered, because that is the column where the value reliably shows up.

Corporate venturing earns its keep as a window on the future, not as a hedge fund with a logo.

The honest limitation: "strategic value" is slippery, and that is exactly how weak programmes hide. A unit that can't point to a specific decision the core business made differently because of a venture is probably generating expensive optionality and calling it strategy. The evidence says strategic value is real and financial out-performance is not the norm, it does not say every programme realises the strategic value. Naming the bet is the start, not the finish; you still have to bank the learning.

Innovation accounting: keeping score before revenue exists

If financial metrics arrive too late to steer a young venture, what do you steer by? This is the problem Eric Ries set out to solve in The Lean Startup (2011) with innovation accounting, his term for measuring progress "when all the metrics typically used in an established company are effectively zero." A new venture can't be judged on revenue or market share because it has neither; judged on a five-year forecast, every team is incentivised to invent a flattering hockey-stick. Innovation accounting replaces the forecast with evidence.

Ries's mechanism has three beats. First, establish a baseline with a minimum viable product, real behaviour from real users, however small. Second, tune the engine: run experiments that move that baseline toward what a viable business would need. Third, decide to pivot or persevere based on whether the numbers are actually moving. The numbers that count are leading indicators in two families: value hypothesis metrics (does using the product genuinely delight people, retention, activation, repeat use?) and growth hypothesis metrics (does that behaviour pull in new users, referral, organic growth?). The discipline is to track actionable metrics that prove or disprove a belief, and to refuse the vanity metrics, cumulative sign-ups, total downloads, that only ever go up and never tell you to stop.

Which points to the practical change: fund ventures in stages against learning milestones, not annual budgets. Each tranche of money buys the answer to a specific question ("will paying customers retain past week four?"), and the next tranche is released only when the evidence clears the bar. That is venture-capital logic, staged, milestone-gated funding, applied inside the corporation, which is the whole point of corporate venturing.

flowchart TD
    A(["State the hypothesis
value or growth?"]) --> B(["Ship an MVP,
set a baseline"])
    B --> C(["Run experiments,
tune the engine"])
    C --> D{"Leading indicators
moving?"}
    D -->|"Yes"| E(["Persevere:
release next tranche"])
    D -->|"No"| F(["Pivot or stop:
change the bet"])
    E --> C
Innovation accounting as a funding loop: money follows evidence, one hypothesis at a time. Leaders Loop

There is a structural counterpart to this. In The Other Side of Innovation (2010), Vijay Govindarajan and Chris Trimble argue that the established business, what they call the performance engine, runs on repeatable, predictable execution, while an innovation initiative runs on the opposite. Hold the venture to the performance engine's metrics and you crush it; let it run with no metrics at all and you can't tell promise from waste. Innovation accounting is the dial that lets the two coexist. One caveat worth saying out loud: leading indicators are easier to game than revenue, and a team can dress up engagement that never converts. Innovation accounting only works if someone with authority is genuinely willing to call "pivot or stop", otherwise it becomes vanity metrics with better vocabulary.

A worked example

The following figures are illustrative, a composite scenario to show the mechanics, not a real company's results.

Picture a mid-sized logistics company. Its CVC unit invests $4m for a minority stake in a route-optimisation startup. Classify the bet first: there's a tight operational link (the startup's software could plug into the company's own fleet), and the objective is strategic, a window on AI routing the parent can't build fast enough alone. That's a driving investment in Chesbrough's terms, which means the scorecard is strategic, not an IRR.

Now apply innovation accounting to the pilot that rides alongside the equity. The hypothesis is a value one: "dispatchers who use the tool will keep using it because it saves real time." Baseline, week 0: dispatchers plan routes manually in about 40 minutes. The leading indicator isn't revenue, it's week-four retention of the tool and minutes saved per route. After the first funded tranche, eight of ten depots are still using it at week four and planning time has dropped to 18 minutes. The metric is moving; persevere, release the next tranche to widen the pilot. Had retention collapsed to two of ten, the same accounting would say pivot, maybe the win is in a different workflow, and that "failure" would be a cheap, fast lesson rather than a buried $4m.

Eighteen months on, the finance review asks "what's the return?" This time there's an answer that fits the bet: the equity mark is secondary; the strategic return is a routing capability now live across the fleet, a de-risked build-vs-buy decision, and a partnership the core business is acting on. That is the column the evidence says CVC reliably fills, and the company can point to it because it decided, up front, which ruler to use.

Frequently asked questions

What's the difference between corporate venturing and corporate venture capital?

Corporate venturing is the broad category, every route an incumbent uses to grow outside its core, from internal startups and accelerators to partnerships, acquisitions and investments. Corporate venture capital is the specific route where the company takes a minority equity stake in an external startup, usually via a dedicated fund or unit. All CVC is corporate venturing; not all corporate venturing is CVC.

Does corporate venture capital actually make money?

On average, the financial case is unproven. The 2023 meta-analysis of 32 studies found a positive link to strategic performance but no significant relationship with financial returns. The most reliable predictor of a programme that survives and adds value is strategic fit between the parent and the startup, not the size of the fund. Treat CVC as a strategic instrument that occasionally also pays; don't run it as a financial fund that occasionally informs strategy.

What is innovation accounting, in one line?

It's how you measure a venture's progress when revenue and market share are still zero: set a baseline with a minimum viable product, track leading indicators of value and growth, and use them to decide whether to pivot or persevere, instead of judging the venture against a forecast it was incentivised to inflate.

Why do so many corporate venture programmes get shut down?

Two patterns recur. They get judged by the parent's standard financial metrics long before any venture could plausibly produce them, and they lack a clear strategic focus, so no one can say what good looks like. Gompers and Lerner found strategically unfocused programmes were markedly less stable and often stopped after only a few investments. Both failure modes are about the wrong scorecard, not the wrong startups.

Can a small company without a fund use any of this?

Yes, innovation accounting needs no fund at all. Any team launching something new can set a baseline, pick one value metric and one growth metric, fund the work in tranches against those, and hold an honest pivot-or-persevere review. That is the transferable half of the toolkit, and it's free.

Related in the Toolkit

Where to go next