Every corporate innovation programme eventually produces a slide of numbers. Ideas submitted, hackathons run, startups met, proofs of concept launched, patents filed, memoranda of understanding signed. The numbers are usually large and rising. The trouble is that almost none of them tells a board whether the company will earn more money in five years than it would have without the programme, and some of them actively mislead, because they reward activity that looks like progress and is not.
Corporate innovation metrics are hard for a structural reason. The outcome that matters, new revenue and profit from new businesses, arrives years after the decisions that caused it. By the time the lagging numbers are visible, the programme's sponsor may have moved on and the budget may have been cut. So programmes measure what they can see early, and what they can see early is mostly effort. The real task is to find early signals that actually predict later value, and to retire the ones that do not.
This article reviews what the research says about how companies measure innovation, separates leading indicators from lagging outcomes and from vanity metrics, and sets out a leading and lagging metric set organised by venture stage that a corporate innovation or venture- building unit can adopt.
What companies actually measure, and why it disappoints
The measurement problem is long-standing. In a 2008 McKinsey Quarterly article, "Leadership and innovation", Joanna Barsh, Marla Capozzi and Jonathan Davidson reported that few senior executives spent time on targets, metrics and budgets for innovation, and that only 30% were held accountable for encouraging innovation through formal targets or metrics in performance reviews. Executives at companies with innovation targets and metrics reported the most confidence in their innovation decisions; overall, 65% of senior executives were only "somewhat", "a little" or "not at all" confident. The authors recommended combining financial targets, such as a goal that 20% of revenue comes from products launched in the past three years, with behavioural ones, such as a target that 25% of ideas come from outside the company.
BCG's 2021 innovation survey shows the gap persisting even as priority rises. Innovation ranked as a top-three priority for 75% of companies, up 10 percentage points in a year and the largest annual increase in BCG's 15 global innovation surveys since 2005. But "just under half" of companies were committed innovators that put real resources behind the priority, and only about a quarter of those scored 80 or above on BCG's innovation-to-impact readiness scale. BCG identified clear C-level ownership as a distinguishing factor between top and underperforming innovators, measured by share of sales from new products and services.
Research from Wharton's Mack Institute, summarised in 2020, explains part of the failure. Saikat Chaudhuri's framework sorts 23 quantifiable innovation metrics into inputs, such as R&D spending and the idea pipeline; process effectiveness measures; and performance outcomes. Output measures, the article notes, are "popular among companies because of their relative ease of tracking", while process measures are harder to quantify but "often are the ones that truly help a firm move innovation forward". George Day, in the same summary, argues that most commonly used metrics are "overly outcomes-oriented" and cannot be reliably attributed to innovation alone. The summary also makes a point that is easy to skip: firms should measure failure rates, because not recording failures suppresses experimentation.
Leading, lagging and vanity: three kinds of corporate innovation metrics
A useful classification separates three kinds of measure.
Lagging outcomes are the results the programme exists to produce: revenue, profit and enterprise value from new businesses, and the share of company revenue they account for. They are unambiguous but slow, and attribution is often contested.
Leading indicators are early measures that, if they move, make the lagging outcomes more likely. The test of a leading indicator is causal plausibility and falsifiability: a reasonable person could explain how it leads to value, and it can go down as well as up. Paying customers at an early stage, retention, a falling cost to acquire a customer and time from gate decision to funding release are examples.
Vanity metrics are measures that look like leading indicators but cannot fall when the programme is failing, or that rise with spending rather than with learning. Ideas submitted, events hosted, startups scouted, proofs of concept started and MoUs signed are the usual examples. They are not worthless as operational counts, but they should never be reported as evidence of progress.
The distinction matters because the vanity metrics are the ones most programmes report. A pipeline of 300 ideas says nothing about whether any of them will become a business. A count of 40 proofs of concept says nothing unless it is paired with how many converted. EY India's survey of more than 200 Indian enterprise leaders for its AIdea of India: Outlook 2026 report shows why conversion is the number that matters: 47% said that more than 10% of their proofs of concept had moved into production. That leaves a majority who could not report even one in ten pilots reaching production, which is how a large count of pilots can coexist with very little value.
What the venture-building evidence says about the lagging numbers
For the lagging outcomes, McKinsey's October 2025 survey of 715 senior managers and C-level executives offers a reference point. Respondents reported that ventures launched in the past five years contribute about 12% of enterprise-wide revenue, and they expect that share to reach about 19% within five years. 61% said their ventures had passed $10 million in annual revenue, up from 45% in 2023, and the average age of ventures above that threshold was about 31 months, down from 38. Only 47% said ventures launched in the past five years had met or exceeded expectations.
Two cautions apply. These are self-reported figures from executives who chose to answer a venture-building survey, so they are likely to overstate typical performance. And they combine very different industries and venture types. They are useful as an order-of-magnitude check on what a mature programme might aim for, not as targets. The more useful lesson is structural: the lagging number that boards care about, share of revenue from new businesses, takes years to move, so the programme must be governed through earlier measures in the meantime.
A leading and lagging metric set by venture stage
The table below is this article's contribution. It organises corporate innovation metrics by the stage of a venture, from thesis to scale, and gives for each stage the leading indicators worth tracking, the lagging outcome that stage should eventually produce, and the vanity metric most often reported in its place. The stage names follow the gate sequence in a separate PanScience Research piece on the corporate venture business case. The metric selection is an analytical framework, not the result of a statistical study; the thresholds a company sets should reflect its own industry and history.
Three features of the table are deliberate. First, every leading indicator can go down. Paying customers can churn, interview evidence can contradict the thesis and gate conversion can fall. That is what makes them informative. Second, the portfolio row includes a kill rate. A programme that never closes a venture is either extraordinarily lucky or not testing anything, and the Mack Institute summary's point about recording failures applies directly. Third, the table includes process measures of the parent's own behaviour, such as time from gate pass to funding release. If ventures pass gates and then wait months for money, the problem is the parent, not the venture, and the metric makes that visible.
Designing the reporting rhythm
Metrics only change behaviour if they are reported to the right people at the right cadence. Three practices help.
Report leading indicators to the venture committee, lagging outcomes to the board. The venture committee that releases tranches needs the stage-specific leading indicators at every gate. The board needs a small number of portfolio-level lagging outcomes once or twice a year, with the leading indicators as context. Showing the board 40 activity numbers invites it to manage the programme through the wrong lens.
Tie metrics to decisions, not rewards alone. The 2008 McKinsey finding that executives with targets and metrics were more confident in their decisions is a reminder that metrics serve decision-making first. Each leading indicator should be linked to a decision it informs, such as continue, pivot or stop. Linking metrics to individual bonuses can be useful at the portfolio level, but tying a venture lead's pay to a single early metric encourages gaming.
Keep vanity metrics, but label them. Counts of ideas and events can be useful for running an operation, and some sponsors will want them. The discipline is to report them under an "activity" heading, separate from "evidence" and "outcomes", so that nobody mistakes one for the other.
Measure the parent as well as the ventures. Most innovation dashboards hold ventures to account and leave the parent's own conduct unmeasured. Yet many of the delays that kill corporate ventures sit with the parent: procurement and security reviews that take longer than the pilot, legal approval of a first customer contract, internal transfer pricing for shared services, or a sponsor who is unavailable for a gate meeting. A short set of parent-side measures, such as days from gate pass to funding release, days to approve a first commercial contract and the number of ventures waiting on an internal decision at month end, turns those delays from anecdotes into data. It also gives the venture committee something to fix that is within its own control.
Choosing targets without inventing benchmarks
Companies often ask what a good number looks like. The honest answer is that public benchmarks are thin and survey-based. The 2008 McKinsey suggestion of 20% of revenue from products launched in the past three years is an example target from the authors, not an industry norm. The 2025 McKinsey survey figure of about 12% of revenue from ventures launched in the past five years is a self-reported average across very different respondents.
A better approach is to set targets from the company's own reverse income statement: the revenue share from new businesses that the strategy requires in a given year, worked back to the number of ventures that must reach scale, the number that must pass each earlier gate given expected kill rates, and the leading indicators each must show. That produces targets the board can defend, because they come from the strategy rather than from a peer comparison that may not apply.
PanScience reports that its portfolio spans more than 24 co-founded ventures across nine verticals. Any organisation running a portfolio of that kind faces the same choice: report what is easy to count, or report what predicts value and accept that some of those numbers will be uncomfortable.
Conclusion
Corporate innovation metrics fail when they measure effort as if it were progress. The research has said this for years: output measures are popular because they are easy, process measures matter more, and most companies still lack clear targets and accountability for innovation. The way out is not more metrics but a better classification of the ones in use: lagging outcomes that define success, leading indicators that can fall as well as rise and therefore predict it, and vanity metrics that should be labelled as activity and kept off the evidence slide.
Organised by venture stage, the result is a metric set that tells a venture committee whether to release the next tranche, tells a board whether the portfolio is on course and exposes the parent's own delays. It will usually produce a smaller, less flattering slide. It will also produce one that a board can use.
Methodology and sources
This article combines published research on innovation measurement with an original stage-based metric framework. The 2008 findings on accountability, confidence and example targets come from Joanna Barsh, Marla Capozzi and Jonathan Davidson, "Leadership and innovation", McKinsey Quarterly, January 2008 (https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our- insights/leadership-and-innovation). The 2021 priority and readiness figures come from BCG, "Overcoming the Innovation Readiness Gap", April 2021 (https://www.bcg.com/publications/2021/most-innovative-companies-overview). The input, process and outcome classification, George Day's view and the point about recording failures come from Ayaka Shimada, "Choosing Innovation Metrics for Success in the Digital Era", Mack Institute for Innovation Management, Wharton, February 2020 (https://mackinstitute.wharton.upenn.edu/2020/innovation- metrics-for-success-digital-era/). Venture revenue figures come from McKinsey, "The way to win in corporate venturing: Serial building and AI", 28 October 2025 (https://mckinsey.com/capabilities/business-building/our-insights/the-way-to-win-in-corporate-venturing-serial- building-and-ai). The proof-of-concept conversion figure is editor-verified from EY India, The AIdea of India: Outlook 2026, Chapter 2 survey findings (16 November 2025, https://www.ey.com/en_in/insights/ai/agentic-ai-india/aidea-survey-findings). PanScience's portfolio figure is company-reported. The leading, lagging and vanity classification and the stage-based metric table are original and illustrative; they are not derived from a statistical study.