This post was originally published on the BE AI READY Substack.
Why obsession with ROI — viewing AI as a cost saver — is blinding us to the real measure of AI as an organizational value creator.
Nearly every conversation I’ve been having about AI in the enterprise eventually comes back to one question: “But what’s the ROI?”
It sounds responsible. Strategic, even.
But the more I sit with executives and hear leaders talk about AI strategy, the more convinced I become… that the ROI question is the wrong question to ask.
Not because ROI isn’t important — it is. But ROI measures efficiency at a specific point, for specific outputs — not value creation. ROI captures how well a process runs; it doesn’t tell you whether you’re running the right process in the first place.
The Real Blind Spot: AI as a Tool
When we measure AI through the lens of ROI, we’re stuck treating it like a tool — something that accelerates existing tasks.
But organizations don’t succeed because of tools; they succeed because those tools are part of systems — where people and processes use those tools to create value. Systems define how decisions are made, how information moves, and how quickly an organization can adapt.
In short, systems define how value is created.
That’s where AI’s true potential lives — not just in automating tasks to get work done faster for the people in the system, but in transforming the systems that shape how that work drives equity for the organization as a whole.
The Mirage of 50-60% Efficiency
We’ve all seen the headlines: “AI boosts productivity by 60%”, “Developers complete tasks twice as fast with Copilot.”
These stats spread fast because they’re easy to digest, and easy to sell.
And to be fair — they’re not necessarily made up.
Studies really do show GitHub Copilot helping developers complete coding tasks up to 55% faster, and generative AI assistants enabling customer service agents to resolve issues about 14% faster.
But here’s the truth hiding in the fine print: those gains are task-specific, not organizational.
When AI is deployed across real workplaces, the numbers flatten… fast.
The Federal Reserve Bank of St. Louis found that among employees actually using generative AI, the average time saved was just 5.4% of total work hours. Across all employees, it’s closer to 1–2%.
Executives I’ve spoken to, from IT to general management all agree, the numbers are much lower than the big headlines tend to report.
That’s not to say 50–60% isn’t happening — it’s just that those numbers are very likely incomplete. It’s measuring productivity at the task level, not the system level. It tells us how much faster the tool works for those people. Not how much better, stronger, or efficient the organization became.
ROI Measures Tools. Value Measures Systems.
ROI has always been about comparing inputs and outputs — how much faster, cheaper, or better something gets.
But organizations don’t run on individual outputs; they run on systems.
Systems that govern how decisions are made, how knowledge flows, how people collaborate, and how feedback loops reinforce learning.
AI’s real power isn’t in doing more of the same thing faster — it’s in changing how the organization learns and decides.
Value as a measure of success, changes the calculus from efficiency as a product of the tool, to productivity as a product of the system built around using that tool.
The Productivity J-Curve
The Productivity J-Curve is an economic term that defines the dip before the lift that happens when a general-purpose technology (like electricity, the internet, or now AI) enters the system.
At first, things get messy. Processes lag. Roles blur. Tools outpace governance. Productivity might even drop.
Sound familiar?
But as organizations restructure — realigning workflows, decision rights, and data foundations — the curve bends upward. That’s when the compounding value starts.
I think we’re entering the dip.
The real vs. expected efficiencies we’re seeing reported (~50% vs ~10%) as AI shifts from an individual tool for efficiency to an organizational tool for equity, is a clear sign we’re already moving downwards.
- Efficiency is the early gain.
- Equity — durable advantage — is the long-term return.
The question now is, how do we shift back up? The answer isn’t “change how we measure success” — it’s to rethink the way we implement AI that drives that change.
From ROI to Organizational Equity
In one of my recent peer groups, a startup CEO and former VC said something that stuck with me:
“As a VC, I never cared about ROI. I cared about equity — because equity is the total value of the business, not the return on a single investment.”
Exactly.
Instead of asking how much time AI saves, we should be asking how much organizational equity it builds.
AI Value = Organizational Equity.
Equity grows when AI helps the organization learn faster, decide smarter, and adapt sooner. That’s value you can’t capture in a timesheet.
The AI Value Scorecard
If we stop tracking “hours saved” and start tracking “capabilities gained,” the metrics start to look different. The Value Scorecard is a framework you can use to identify common areas where AI helps organizations grow smarter — not just faster. This is just a starting point that should get you started on determining what drives value for your organization.
This is the same framework we use in the Cognitive Operating Model to measure organizational success with AI.
The Shift That Actually Matters
AI does create ROI. People are definitely getting things done faster, and arguably, their able to do more because of that time savings. But that story isn’t the one that really matters at the organizational level, because you can’t roll up hours saved into real organizational gains — 1+1 may only add up to 1.2, not 2.
Sure, it looks good on a balance sheet to say you can replace 10 FTE’s with an AI agent that costs 1/4 of their salaries. But those cost savings will quickly get eaten up in other places. What really matters long term? How are those gains reinvested into learning, agility, and innovation — the things that build equity?
AI excels for individuals as a tool.
But organizations don’t run on tools.
They run on systems. And systems are where equity is built.
The organizations that will win aren’t the ones saving the most hours. They’re the ones using those hours to build better systems — systems that learn, decide, and adapt faster than the competition.
That could be realized as cost savings, but it definitely shows up as compound growth.
Why Microsoft Did This, and What It Means For Leaders and Their Orgs
Let’s be perfectly frank — this wasn’t Microsoft being generous. It’s a strategic play.
- Competition: ChatGPT and Gemini already give away powerful free tiers.
- Seeding demand: Some workers will inevitably run up against the limits of free Copilot and request licenses from their employers.
- Lock-in: Once Copilot is part of your daily workflow, you’re less likely to normalize to external AI tools, especially if they can’t incorporate email and Teams conversations.
For leaders, Free Copilot for all changes the starting point of the AI conversation:
- Before: Copilot was a purchase decision. Leaders could hold off based on cost, ROI, and security concerns.
- Now: AI is becoming a commodity, accessible to every worker — for free.
As executives and IT leaders come to terms with the reality of AI infiltrating every aspect of work, it’s becoming clear that avoidance as a strategy is neither effective nor practical.
Being AI ready means establishing governance, training, and usage policies before the “lite” experience creates uncontrolled habits.
Making AI a Platform for Organizational Success
Microsoft’s move to make Copilot accessible for everyone, alongside the integrations of AI into Teams, is all about making their platform all about productivity.
But more AI ≠ more productivity.
This is exactly the moment where the Cognitive Operating Model (COM) becomes not only useful — but a critical framework for ensuring AI is being implemented and used accountably, reasonably, and effectively. It’s a system we’ve developed to help organizations:
Map workflows where AI creates measurable value.
Design delegation models for humans + AI.
Build feedback loops for accuracy, trust, and compliance.
Without a framework, universal AI risks chaos. With COM, it becomes a foundation for scaling responsibly. Send me a chat and ask me about the COM framework anytime!