OpenAI CFO's Four-Question Framework for Measuring AI ROI

A New Metric for AI Value

As finance organizations pour billions into artificial intelligence, many CFOs are discovering that traditional metrics—model specifications, vendor promises, cost per token—fail to capture whether AI investments actually deliver business value. OpenAI CFO Sarah Friar is proposing a different approach: measure "useful intelligence per dollar."

In a recent blog post, Friar outlined four questions she uses as a scorecard for evaluating AI deployments. The framework pushes finance leaders to identify which work truly matters to their business, calculate the fully loaded cost when AI performs that work, and test whether employees can reliably build on AI outputs rather than redoing them. The central question: Is the value of AI-completed work compounding faster than the cost to produce it?

Why it matters

With 83% of CFOs planning AI budget increases above 15% over the next two years—and 42% expecting increases above 30%—finance leaders need concrete frameworks to separate productive investments from expensive experiments. Yet only 31% of CFOs currently rate AI outcomes in finance as strongly positive, according to Bain & Company research. The gap between organizations that have successfully scaled AI and those still struggling is widening, making effective measurement frameworks increasingly critical for competitive positioning.

The Investment Surge Continues

Finance executives are accelerating AI spending despite mixed early results. A Bain & Company survey found that 56% of senior finance leaders are increasing enterprise-wide AI investment by more than 15% this year. The research, first reported by Fortune, suggests CFOs are doubling down not because returns have been spectacular, but because the competitive disadvantage of falling behind has become too significant to ignore.

Friar's framework aims to help finance teams move beyond hype and establish whether AI deployments are creating measurable value. Her questions focus on defining meaningful work, quantifying total costs, validating output quality, and proving that value compounds faster than expenses—or reallocating capital before enthusiasm turns into financial drag.

Finance Leaders Seek Viable Scorecards

As CFOs increasingly take responsibility for steering AI value creation across their organizations, many are still searching for practical measurement approaches. Friar's four-question framework offers a concise checklist that finance teams can adapt to their own pilots and programs, providing a structure for evaluating whether AI investments justify their growing budgets.

The framework represents a shift from technology-centric metrics to business-outcome measures, reflecting the maturation of AI from experimental technology to operational tool that must demonstrate clear return on investment.

These details were first reported by Fortune.