🔍 Why Measure Innovation ROI—and the Pitfalls
Measure innovation ROI and at the same time essential and dangerous. Essential because without numbers, the program loses credibility, budget, and its sponsor in every round of cuts. Dangerous because applying the wrong metrics at the wrong time can kill bets worth billions. The classic trap: demanding the same ROI standard from a 30-day H3 experiment as you would from a mature business line with 10 years of history.
📊 The Measurement Paradox
- • 76% of CFOs report difficulty justifying innovation investments without short-term returns (Deloitte CFO Survey, 2024)
- • 84% of radical innovations that generated more than US$ 1 bi would have been cut if evaluated using traditional financial metrics in their first 2 years (HBR, “The Innovator's Dilemma at Work”)
- • Amazon Web Services: generated losses for 7 years before becoming Amazon’s most profitable business — with a 30% operating margin vs. 3% for e-commerce
- • Companies that require a positive ROI in less than 18 months for all projects innovate 2.8x less than competitors with a portfolio perspective (BCG Innovation Report, 2023)
🎯 The Right Question
The mistake isn’t measuring innovation ROI. The mistake is using A SINGLE metric to evaluate THREE fundamentally different types of the bet. Each innovation horizon requires a distinct language of value:
H1 — Core
Traditional financial ROI is appropriate. Payback, NPV, IRR. The business is mature, and the cash flow is predictable.
H2 — Adjacent
A mix of financial and learning metrics. Acquisition cost, retention, emerging unit economics.
H3 — Transformational
Innovation accounting: the ROI is validated learning. Financial returns only show up years later.
🎁 Types of Returns: Financial, Strategic, Learning, and Option Value
Traditional financial language recognizes only direct monetary returns. But innovation generates four value categories — and ignoring the last three is like judging an iceberg by what appears above the water. Each category has its own metrics and a different time window.
💵 Financial Return
The easiest to measure, the easiest to manipulate. Incremental revenue from new products, cost reductions through automation, margins on new services. A good indicator for H1, misleading for H2, and useless for H3 in the early years.
Metrics: NPV, IRR, payback, revenue from new products as % of total (McKinsey target: 25% in 3 years)
⚖️ Strategic Return
Competitive positioning value. Capture a segment before competitors, build a data moat, acquire critical capabilities. Nubank invested years in proprietary technology without immediate ROI — the strategic return was a structural cost advantage that enabled 10x growth.
Metrics: market share, NPS relative to the market, time for a competitor to copy, intangible assets (patents, data, brand)
🧠 Learning Return
The most undervalued asset. Every experiment that fails but generates insight reduces the cost of future experiments. Organizations that systematize learning reach the right product 40-60% faster on the second attempt (MIT Sloan Management Review). Toyota calls this "accumulated knowledge"—the main asset of the Toyota Production System.
Metrics: number of hypotheses tested and documented, iteration speed, reuse of lessons in new projects
🎭 Option Return
Real options theory applied. An H3 experiment costing R$ 500 mil doesn’t cost R$ 500 mil: it buys the right to scale a R$ 500 mi business if the hypothesis is validated. Failure costs R$ 500 mil. Success opens up an option valued at orders of magnitude more. Google X (moonshot factory) operates exactly this way.
Metrics: expected option value (probability x market size), experiment cost as a % of the potential market
📈 Real Options and Portfolio View
The theory of real options (Real Options Analysis — ROA), developed by Stewart Myers at MIT in 1977 and popularized by Aswath Damodaran, applies the logic of financial options to strategic decisions. The premise: in highly uncertain environments, the value of a decision isn't in its expected cash flow — it's in the the flexibility to act in the future with information that doesn’t exist today.
📊 The 3 Types of Options in Innovation
Growth Option
A small investment today that opens the possibility of a large investment tomorrow. E.g.: Embraer invested R$ 80 mi in aviation biofuel R&D over 4 years — with no immediate ROI. In 2027, when regulations take effect, the company will already have the capacity in place. Competitors will pay 5x more to build from scratch under regulatory pressure.
Deferral Option
The right to wait until you have more information before committing capital. A pilot with 200 users isn’t “taking too long”—it’s buying information cheaply before scaling at a high cost. Magazine Luiza ran regional marketplace pilots for 18 months before scaling nationally, avoiding a logistics mistake that cost a competitor R$ 2 bi.
Abandonment Option
The right to stop before losing more. Having clear kill criteria is part of the value—it reduces the portfolio's average cost by preventing "zombie projects" that consume resources without delivering. BCG estimates that 30% of innovation projects at large corporations are zombies kept alive by internal politics, not merit.
💡 The Logic of the Innovation Portfolio
No venture capitalist expects EVERY investment to pay off. The model is: 6 bets fail, 2 return the capital, 1 returns 3x, and 1 returns 30x. The portfolio’s return exceeds any individual bet. Innovative companies should think the same way when managing H3: the portfolio’s expected value is positive even when most bets fail, as long as the cost of each experiment is calibrated to the market’s potential size.
Practical formula: The cost of an experiment should be < 1% of the addressable market. If the market is worth R$ 1 bi, spending up to R$ 10 mi to validate the hypothesis is economically rational.
⚠️ Pitfall: Correlating Investment Size with Importance
Managers trained in traditional projects automatically associate "little money = unimportant." In H3 innovation, the smallest experiment can open up the biggest market. Amazon invested US$ 6 mi in the Kindle MVP — a product that redefined the book market and generated US$ 3 bi in revenue. Protect small experiments from being cut because they seem insignificant in budget meetings.
📋 Innovation Accounting — Learning Metrics (Ries)
Eric Ries, in “The Lean Startup” (2011) and its follow-up book “The Startup Way” (2017), formalized the concept of innovation accounting: a parallel accounting system designed to measure progress under conditions of extreme uncertainty, where traditional financial metrics do not apply because the product, market, and business model are still being discovered.
📈 The 3 Levels of Innovation Accounting
Level 1: Baseline
Establish the real starting point—not the plan. What is the initial engagement? What is the conversion rate without optimization? What is the cost per experiment? Ries calls these “baseline metrics”: the honest point from which the startup/project stands today.
Level 2: Growth Engine Metrics
Identify which of the three levers is the business's growth engine: viral (viral coefficient > 1.0 = each user brings in more than one), paid (LTV/CAC > 3x in 12 months), stickie (retention > 40% in month 3). Measuring only the relevant engine eliminates “vanity metrics.”
Level 3: Pivot or Persevere Metrics
The most expensive decision in innovation: when to stop optimizing what isn’t working and change direction. Innovation accounting defines thresholds: if the key metric doesn’t improve by X% after Y iteration cycles, the team presents evidence to the board to decide whether to pivot.
✓ Actionable Metrics
- ✓Activation rate (% who reach the "aha moment")
- ✓Cohort retention (month 1, 3, 6)
- ✓Willingness to pay (genuine willingness to pay)
- ✓Transactional NPS (not annual)
- ✓Cost per validated learning
✗ Vanity Metrics
- ✗Total number of registered users
- ✗Downloads/installations (no activation)
- ✗Page views without engagement
- ✗Number of ideas submitted (without screening)
- ✗Revenue from a single customer (no recurring revenue)
👤 How to Communicate Value to the Board and Executive Team
The board speaks the language of risk and return. The innovation team speaks the language of learning and experimentation. This linguistic difference kills more innovation programs than any technical failure. Translation is the most critical skill for an innovation leader. It's not about hiding uncertainty — it's about framing uncertainty in terms the board recognizes as responsible risk management.
🎯 The Innovation Board Dashboard
1. Portfolio by Horizon
How many projects are in H1/H2/H3, and how much budget is allocated to each. Shows strategic balance—in language CFOs and board members understand.
2. Future Value Pipeline
Revenue projection for new products in 3/5 years, with confidence intervals. No false precision: "we believe this portfolio can generate R$ 200-500 million in 5 years".
3. Speed Metrics
How many experiments we run, how many hypotheses we validate or invalidate, average cycle time. Shows that the program is functioning as a learning machine.
4. Market Comparison
Investment in innovation as a percentage of revenue vs. the industry benchmark. If the industry invests 4% and the company invests 1%, the board understands the risk of underinvestment.
💡 The 3 Levels of Narrative Framework
Rita McGrath (Columbia Business School) recommends three levels of narrative for communicating innovation to the board, depending on the time horizon:
🎮 Communication Mistakes with the Board
- ✗ Oversell: promise ROI that doesn't exist. When it doesn't materialize, the board cuts everything — including the projects that would have worked.
- ✗ Untranslated jargon: using "pivot," "MVP," "product-market fit" without context. The board interprets it as imprecision, not methodology.
- ✗ Reporting only failures without learning: “the pilot didn’t work,” without “here’s what we learned and how it changes our next bet,” sounds like incompetence, not systematic experimentation.
- ✓ Correct structure: Cost of the entire portfolio > Hypotheses tested > What we validated > What we invalidated > How this changes the direction > Next investment cycle.
⚠️ Classic Mistakes: Demanding ROI Too Soon and Looking Only at Short Payback
Most innovations that fail don't fail because of a lack of ideas or execution — they fail because they were dead before reaching the inflection point. The two most destructive mistakes are demanding financial returns before product-market fit, and using short-term payback as the sole criterion for allocating capital to exploratory innovation.
📈 The Innovation J-Curve
Every H2/H3 innovation project follows a J-curve: increasing investment with negative returns at first, an inflection point when the hypothesis is validated, and accelerated growth afterward. Those who cut the project during the negative phase of the curve — exactly when learning is happening — never reach the positive side.
Phase 1: Burn
0–18 months. Investment with no visible return. Learning curve accumulating. This is where 70% of programs are cut.
Phase 2: Inflection
Validated hypothesis, product-market fit found. Retention and growth metrics are starting to confirm the model.
Phase 3: Scale
Exponential growth. Financial ROI becomes visible. This is when the late movers want to join—but the early mover already has a structural advantage.
✗ The 5 Most Costly Mistakes
- ✗ Payback < 24 months for H3: is like saying you want to harvest before you plant. No disruptive innovation pays for itself in 2 years.
- ✗ Same discount rate for H1 and H3: ignoring the different risk profiles of each horizon distorts NPV and penalizes long-term bets.
- ✗ Requiring a complete business case at Gate 0: in H3, the business case is fiction. Who could have predicted the streaming market in 2003?
- ✗ Cut H3 during financial crises: just when competitors are vulnerable and the cost of talent is low—that's exactly when you should invest.
- ✗ Treat all innovation R$ as equal: R$ 1 mi in H1 has completely different risk and return from R$ 1 mi in H3.
✓ What to Do Instead
- ✓ Define success metrics by horizon before you start—not at the time of evaluation.
- ✓ Use clear “kill criteria”: the project dies IF it does not reach X within Y months. Remove politics from the decision.
- ✓ Separate budgets by horizon with protection from operational cuts—the H3 budget doesn’t fund an H1 crisis.
- ✓ Review the portfolio quarterly, not annually—lets you reallocate to what is working more quickly.
- ✓ Celebrate experiments that invalidated hypotheses — negative learning is just as valuable as positive learning for reducing portfolio risk.
💡 The Opportunity Cost of Underinvestment
Kodak invented the digital camera in 1975—10 years before any competitor. It cut investment because it threatened the film business. In 2012, it went bankrupt. Nokia had the world’s first smartphone (Communicator, 1996). It cut software investment because it didn’t “fit the hardware ROI.” In 2013, it sold its mobile phone division to Microsoft for a fraction of its value.
The real cost of underinvesting in H3 isn't what you spent—it's what your competitor gained while you waited for payback.
📋 Module Summary
Next Module:
3.8 - Scaling Innovation — how to take a successful pilot across the organization without losing its essence