As we enter the latter half of 2026, the AI conversation has shifted from implementation to ROI. Ninety percent of organizations now report getting some value from AI tools. Only 45% are getting a great deal of it. That gap, which is the distance between “something” and “substantial,” is where the entire AI economy is currently parked.
Co-Founder & CEO at RoAI Institute.
In Harvard Business Review earlier this year, my team and I published findings from a survey (link 1) of 1,006 C-level executives across 32 industries and 11 countries. Several factors separate the organizations getting real returns from those merely getting by. Most prominently, 55% of executives cite unready data as an inhibitor, and 47% cite the absence of a repeatable value framework.
But one variable did more to predict high value than any other: how organizations measure and report what their AI is actually worth.
We mapped that into a six-stage Economic Maturity Model. Stage 0 is unmeasured pilots. Stage 5 is formal reporting of AI value to boards, investors, and public markets. The share of organizations achieving high value climbs from 4% at Stage 0 to 85% at Stage 5, a 20x difference, and the single largest effect size in the entire study.
This is a financial measurement maturity model that measures not how well an organization builds AI, but how it measures what AI delivers.
Two cliffs, not a smooth climb
The journey between Stage 0 and Stage 5 is not linear. Plot high-value achievement against stage and two cliffs emerge, with plateaus in between.
The first cliff sits between Stage 2 and Stage 3, a 24-point jump. This is where organizations stop relying on pre-launch business cases alone and measure outcomes after deployment.
The second cliff sits between Stage 4 and Stage 5, a 27-point jump, the largest in the model. This is where organizations stop reporting AI value privately and start reporting it formally to boards and to markets. Everything between those two moves is incremental. The cliffs are where the value lives.
The 41-point trap
Thirty percent of organizations in our sample are stuck at Stage 3, with a median of six years in AI. The reason they stop is that Stage 3 feels like winning. Forty-four percent of Stage 3 organizations report a great deal of value from AI, one-third say they have substantial positive ROI, and 70% plan to increase AI investment next year.
These numbers sound like success, but at Stage 4, 58% see high value, and at stage 5, it is 85%. Stage 3 organizations are leaving 41 percentage points on the table but not feeling it, because everything is trending up from where they started.
Why do they stay stuck? Stage 3 organizations are measuring, but can’t combine what they measure into a value statement that lands with the board and the C-suite. Finance uses IRR and NPV.
Operations uses efficiency and cycle time. Risk and compliance uses incident reduction. Customer teams use CSAT and retention. Every number is real and defensible, but when the CEO or board asks, “What is our total AI value?” the answer is a shrug or four disconnected numbers from four different functions. Meanwhile the portfolios keep growing, and the translation gap scales with them.
The missing seat at the table
This brings me to the most actionable finding in our research. When the CFO and the finance function are involved in achieving and certifying AI value by partnering with the CIO, CTO, or chief data and AI officer, 76% of organizations achieve high value from AI. When the CIO or CTO carries that accountability alone, the rate drops to 53%. When functional executives carry it, 32%. Yet today, only 2% of organizations involve the CFO in this role.
The point is not to put the CFO in charge of AI, but to provide institutional credibility. Finance’s existing job is to translate different kinds of value into one common unit, and its certification of a number makes the rest of the organization believe it. The unit-CFO validation step is what makes the aggregate credible to the board, to investors and to the market.
This is also what unlocks the second cliff. As organizations move from Stage 4 to Stage 5, three things change at once, none of them technical. CIO and CTO involvement in AI value nearly doubles, from 20% to 39%. The inhibitor “our leaders don’t understand AI value” drops from 40% to 18%, because formal reporting forces executives to defend the number in public.
And investment conviction follows the proof. Just over half (52%) of Stage 5 organizations plan to substantially increase AI investment next year, which is ten times the rate at Stage 3.
Stage 4 organizations know their AI number. Stage 5 organizations are accountable for it. That shift, from private knowledge to public accountability, is what unlocks the final 27 points.
Three questions for your leadership team this week
If you take nothing else from this research, take these three:
One. Can we quantify our total AI value today, as a single number?
Two. Who owns that number — not the AI team, but an executive accountable directly to the CEO, with finance at the table?
Three. What would it take to put AI value on our board agenda as a recurring item?
If you cannot answer the first question, you are at Stage 3 or below, the same company as 62% of organizations globally. The good news is the path forward is not a technology problem, but a management decision. And it can start this quarter.
We've featured the best AI website builder.
This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit