Cluster 2 of 10
From 150 Innovation Leaders
You’re Measuring Innovation Wrong
Companies track more KPIs than ever. Innovation readiness just hit an all-time low. The metrics aren’t broken. You’re measuring the wrong things entirely.
83% Priority
say innovation is a top-3 priority
vs.
3% Ready
qualify as innovation-ready
The Measurement Paradox
I’ve run innovation programs with 35+ Fortune 100 companies. Almost every one of them had a beautiful innovation dashboard. Almost none of them could tell me whether their innovation efforts were actually working.
BCG’s 2024 Most Innovative Companies report found that 83% of senior executives call innovation a top-three priority. But only 3% of their companies qualify as innovation-ready. Down from 20% in 2022.
The BCG data matches what I see on the ground: innovation spending keeps climbing, but the share of companies generating meaningful returns from it keeps falling. They added dashboards, hired analysts, built reporting layers. The result? More measurement, less actual experimentation.
I asked. They answered.
I asked 150 innovation leaders from 100+ companies to name their biggest innovation challenge. Measurement was cluster #2. The same pain repeated 15 times: “We can’t prove innovation ROI, so leadership won’t fund it.”
0 Leaders
0 Cited Measurement
0 Companies
Three Metrics You’re Tracking (and What to Track Instead)
01 Cross-Functional Collision Rate
Stop Counting Ideas. Start Counting Collisions.
When I asked 150 innovation leaders at the summit what their biggest challenge was, the most repeated complaint wasn’t about creativity. It was this: “We have plenty of ideas. We can’t get the right people working on them together.” I’ve heard this exact sentence at pharmaceutical companies, defense contractors, and consumer brands. It’s a relationship infrastructure problem.
McKinsey research shows companies emphasizing cross-functional teams are 25% more productive and 30% more efficient at implementing new ideas. The metric: how many unique cross-functional pairs form each quarter?
Majority: track idea volume Gap Minority: track who connects
Family Story
📌 The Post-it Notes Problem
My grandpa worked on the Post-it Notes project at 3M. The most successful consumer product in the company’s history almost never happened. Not because the idea was bad. Because the hierarchy said no.
1968 Spencer Silver invents a low-tack adhesive at 3M. Nobody knows what to do with it. Management says it’s a solution looking for a problem.
1974 Art Fry, a different 3M scientist, uses Silver’s adhesive to bookmark his church hymnal. The Post-it concept is born. Not in a boardroom. At church.
1974-1977
Three years of internal rejection. The hierarchy keeps saying no. Too small a market. Can’t prove ROI. No clear category. The formal approval process kills it repeatedly.
1977 Fry starts giving free samples to secretaries at 3M. Word spreads through hallways, churches, and softball fields. Informal networks do what formal hierarchy couldn’t: build demand.
1980 National launch. 12 years from invention to market. The product that “couldn’t prove ROI” now generates billions. It was approved not by the hierarchy, but by the people who actually used it.
“Post-it Notes never would have gotten approved if it wasn’t for the collisions of people at the churches and softball fields outside the office. It was always just ‘no’ in the hierarchy. They were all siloed.” My grandpa, who was on the Post-it Notes project at 3M
That quote shaped how I think about innovation measurement. It was 1980, but the pattern hasn’t changed. I see it in every Fortune 100 innovation program I run. The metric we should have been tracking wasn’t “did management approve it?” It was “are the right people colliding outside the formal structure?” Fry found Silver’s adhesive because they existed in the same informal network, not because 3M’s hierarchy connected them.
02 Experiment Velocity
Your Innovation Budget Isn’t a Budget. It’s a Bet Portfolio.
Leader after leader at the summit told me the same thing: “Leadership wants ROI before we’ve even run the experiment.” I’ve sat in those budget meetings. This kills innovation at the root.
Here’s the joke that isn’t a joke: to get something innovative approved at most large companies, it doesn’t matter if the idea is brilliant. It has to survive layers of possible “no” before reaching anyone with decision authority. And there’s no intelligent system finding the person who actually needs it to make the product profitable.
👤 Team Lead Possible No
👤 Department Manager Possible No
👤 Senior Manager Possible No
👤 Director Possible No
👤 VP of Innovation Possible No
👤 SVP Possible No
👤 Chief Innovation Officer Possible No
👤 CFO (Budget Gate) Possible No
👤 CEO Possible No
✅ Approved 1 Yes Needed
9 layers of “possible no” to reach 1 “yes.” Wharton research shows that hierarchy beyond ~9 levels actively stalls innovation.
Amazon
1,976 experiments per year. Less than 50% improve their target metric. Bezos: “Given a 10% chance of 100x payoff, take that bet every time.” Source
3M 169
new products launched in 2024 (32% increase YoY). Their Vitality Index demands 30% of revenue from products less than 5 years old. Source
Prediction Markets
What If Innovation Had a Prediction Market?
This is the part that excites me most. Prediction markets like Polymarket just hit $9B in trading volume. They predicted 49 of 50 states in the 2024 election. Traditional polls missed by 3-5 points. Why? Because markets aggregate distributed knowledge better than hierarchies.
$9B Polymarket 2024 volume (up from $73M in 2023)
The Block, 2025
25% reduction in forecast error vs. expert predictions when companies use internal markets
Google, Ford, Intel research
Google, Intel, HP, and Ford have all run internal prediction markets. Google’s markets predicted product demand more accurately than official forecasts. The pattern I keep seeing in every company I work with: the crowd of employees who actually build and use the products knows more than the executive who approves the budget. So why do we still measure innovation by asking executives to approve it?
03 Knowledge Compound Rate
Innovation That Doesn’t Compound Is Just Activity.
Patents filed. Projects completed. Ideas in the pipeline. I’ve watched companies celebrate these numbers in quarterly reviews while their actual innovation capacity erodes. These metrics share a fatal flaw: they treat innovation as discrete events. File a patent. Close a project. Move on. But the companies that actually innovate treat knowledge as compound interest.
The VC Portfolio Lesson
Venture capital figured this out decades ago. Analyzing 11,350 startups from 1986 to 2018: 60% failed to return investment. 27% were complete write-offs. Only 1.1% returned an entire fund. The top 6% of deals generated 60% of all returns.
VCs don’t measure each bet by individual ROI. They measure portfolio learning velocity: how fast does each failure improve the next bet? That’s the power law at work. Companies do the opposite. They treat every innovation project as a standalone P&L, killing anything that doesn’t show returns in 18 months.
27% Dead
33% Break Even
34% Returns
6% ★
VC Portfolio Distribution: 6% of deals generate 60% of all returns
Andrej Karpathy demonstrated compound knowledge at the individual level. He published 100+ articles totaling 400,000+ words, building a living wiki on LLM research. Each article compounds on the ones before. The knowledge doesn’t just accumulate. It connects. That’s the metric. Not “how much did we produce?” but “how much of what we produced builds on what we already know?”
| What You Track | What Actually Compounds | Why It Matters |
|---|---|---|
| Patents filed per year | Revenue from products <5 years old | Measures whether knowledge reaches the market |
| Ideas in pipeline | Experiments that build on prior results | Measures knowledge reuse, not just creation |
| Innovation budget spent | Learning velocity (experiments per dollar) | Measures efficiency of learning, not spending |
| Project completion rate | Portfolio learning rate (a la VC power law) | Measures whether failures improve future bets |
Frequently Asked Things
Won’t leadership demand ROI metrics anyway?
Yes. Run a dual dashboard: leading indicators (collision rate, experiment velocity, compound rate) for the innovation team, and lagging indicators (revenue, margin) for the board. The leading indicators predict the lagging ones. Once leadership sees the correlation, they’ll trust the process.
Could prediction markets work inside our company?
Google, Intel, HP, and Ford have all tested them. The results: 25% lower forecast error than expert predictions. The barrier isn’t technical. It’s cultural. Prediction markets work by aggregating distributed knowledge, which means admitting the crowd sometimes knows more than the executive. That’s a harder sell than any technology.
How do you actually measure “cross-functional collisions”?
Count unique cross-department pairs on project teams, in workshops, or in structured collaboration sessions. Track who worked with whom, on what, and was it their first time collaborating? I built a relationship intelligence layer across 18,752 people that does exactly this. Even a simple spreadsheet works to start.
What’s a good experiment velocity target?
Amazon runs about 38 per week. You’re not Amazon. Start with one experiment per team per sprint. The number matters less than the trend. Are you running more experiments this quarter than last? If yes, your velocity is improving. If the number is zero, that tells you everything about your real appetite for innovation.
What’s the first metric I should change?
Experiment velocity. Easiest to measure (count experiments per quarter), hardest to fake (you either ran one or you didn’t), and most culturally revealing. If the number is zero, stop reading blog posts about innovation and go run an experiment.
What if your innovation dashboard showed collision rates, experiment counts, and compound learning instead of idea submissions, patent filings, and project ROI?
My grandpa’s Post-it Notes took 12 years to get through the hierarchy. Prediction markets outperform expert panels by 25%. Amazon’s failed experiments generate more value than most companies’ successful ones. After running innovation programs with dozens of the world’s largest companies, I’m convinced: the measurement system you choose shapes the innovation culture you get.
Cluster 2 of 10
Ready to Measure What Matters?
Second in a 10-part series from 150 innovation leaders. Each cluster reveals a different pattern.
