AI ROI framework: Link Spend to EBIT in Four Steps

AI ROI framework for CFOs and CEOs: a four-step impact-chaining model translates workflow-level AI activity into board-ready financial terms.

Nine in ten companies are spending on AI. Only four in ten trace any EBIT movement from it. The gap is not a technology problem — it is an attribution problem, and it is costing boards their ability to make a rational next budget decision.

The activity trap executives keep falling into

A February 2026 NBER survey of 6,000 executives across the U.S., U.K., Germany, and Australia found 90% report no detectable AI impact on productivity or employment over the past three years. Two-thirds of those same executives use AI personally for 1.5 hours each week. They are not disengaged. Active users simply cannot connect what they do to what the income statement shows.

CMR Berkeley's November 2025 analysis offers a competing diagnosis: the failure is an accountability problem, not a measurement one. Executives who treat AI as IT's job will not use a better measurement model, because using it would force them to own outcomes they have structurally avoided. The argument is worth taking seriously. And it does not explain the NBER data. If disengagement drove the failure rate, engaged executives would show better outcomes. They do not.

Chamarthi, an enterprise transformation leader who published work on this in 2026, puts the failure more precisely: AI does not produce financial results when it is not tied to P&L movement. This is not a governance observation. It is an attribution one.

What a working AI ROI framework chains together

Campbell Harvey at Duke's Fuqua School of Business noted in 2026 executives consistently underestimate AI's reach into coding and strategic analysis — and look for its value in headcount reduction, where it appears last. Accountants built standard financial reporting to track labor and capital. AI value accumulates in workflow speed and decision quality — neither of which show up in a payroll line.

The four-step impact-chaining model closes gap by forcing a specific sequence. Start with a P&L line you own — revenue or operating cost. Identify the workflow moves it. Measure what AI changes in workflow: time per task and throughput volume. Then convert the workflow change into the financial unit the board tracks.

A contract review team cuts average review time from four hours to 45 minutes. Multiply it by deal volume and show the board how many more contracts close per quarter without adding headcount. The chain is what makes it land. Without it, the AI spend sits in the technology budget with no owner and no outcome.

Where the model breaks without executive ownership

McKinsey's 2026 analysis found the four in ten companies tracing EBIT impact share a common pattern: a visible link between AI activity and a financial outcome someone is accountable for. CMR Berkeley is right accountability matters. The attribution model does not replace ownership — it makes ownership legible. A CFO who builds the chain and then hands it to a CTO with no P&L responsibility has not solved the problem.

I have watched organizations spend eight months building AI measurement dashboards no one in the C-suite reads. The dashboards report to a VP of Data who reports to a CIO who has no budget authority over the business units generating the outcomes. Nobody pulls it.

The four steps, stated plainly

Pick one P&L line. Not AI broadly — one line. Identify the specific workflow feeding that line. Measure AI's effect on workflow in units the workflow already tracks. Attach those units to the financial metric using the conversion rate your finance team already accepts.

Gartner's 2026 research confirms AI projects without this kind of governance and integration fail to deliver sustained value. The word "integration" here does not mean technical integration. Finance teams integrate the measurement into a reporting cycle where someone's performance depends on the number.

Global AI spend exceeded $300 billion in 2026. Torsten Slok, Apollo's chief economist, noted this year AI's impact stays invisible in profit margins outside the Magnificent Seven. This is not a patience problem. Companies showing EBIT movement from AI are not waiting longer — they started with the financial outcome and built the attribution chain backward from it.