A new approach aims to help companies track AI’s real business impact. OpenAI’s CFO details a framework focused on outcomes, cost, and reliability. The method targets enterprise needs as AI adoption grows.
Media and publishing leaders seeking to justify AI spending now have a new measurement tool from OpenAI. Sarah Friar, OpenAI’s CFO, has introduced a framework designed to help organizations evaluate the true value of their AI investments by focusing on business outcomes rather than just usage metrics.
According to Friar, the framework centers on tracking the number of successful tasks completed by AI systems and calculating the total cost required to achieve those results. She explained that as AI models become more capable and efficient, companies can accomplish more valuable work at a lower cost, freeing up staff to focus on judgment and creativity.
The core metric, described as “Useful Intelligence per Dollar,” is intended to answer four key questions: whether AI is performing meaningful work, the cost per successful task, the reliability of results, and whether increased AI usage delivers greater value for each dollar spent. Friar noted that costs can vary widely depending on the complexity of the task, such as coding, research, or financial analysis, which may require more computing power and human oversight.
OpenAI outlined steps for businesses to calculate this metric, including factoring in not just the price and compute used, but also employee time, human review, retries, and rework. The company emphasized that the lowest price per token does not always translate to the lowest cost per outcome, as more advanced models may deliver correct results in fewer attempts, reducing overall expenses.
Industry voices have responded positively to the focus on outcomes and reliability. Management consultant Xavier Tsang commented that measuring AI by work accomplished brings the discussion closer to real enterprise value, and suggested that future scorecards could also track human intervention and the cost of unsuccessful outcomes. This approach, he said, would help organizations understand not just how much work AI completes, but also how safely and sustainably that value is created.
As companies look for more precise ways to assess AI’s business impact, some are also exploring new metrics for AI visibility and effectiveness. For example, publishers are increasingly interested in how their presence in AI-powered answer engines can attract advertisers, as discussed in a recent analysis of AI visibility as a key metric for publishers.
OpenAI, founded in 2015, has rapidly become a leading force in artificial intelligence research and deployment. The company’s flagship products, including ChatGPT and the OpenAI API, have seen widespread adoption across industries. As of 2026, OpenAI employs over 1,000 people and has secured billions in funding, positioning itself as a central player in the enterprise AI market.
