OpenAI представила scorecard для оценки ROI от ИИ: четыре метрики вместо подсчёта токенов

Sarah Friar, OpenAI's Chief Financial Officer, published "the scorecard for the AI era" on the company's blog on July 17, 2026 — a system of four metrics for assessing the return on investment of enterprise artificial intelligence: useful work, cost per successful task, reliability, and return on compute.

Which metrics make up the scorecard

OpenAI's scorecard includes four metrics that together answer what Friar calls the key economic question for CFOs: is the value of the work AI performs growing faster than the cost of producing it. She calls the resulting formula "useful intelligence per dollar."

• Useful work — AI must complete meaningful tasks: resolved customer requests, deployed code, reviewed contracts, not just "active usage"
• Cost per successful task — the full price of a result: AI expenses, retries, and the human time spent on review, not just the token price list
• Reliability — how consistently the system delivers results that can be trusted without constant oversight
• Return on compute — whether every dollar invested in AI delivers ever more value as usage grows
• Publication — July 17, 2026, OpenAI blog, an op-ed column by CFO Sarah Friar

Why AI cannot be measured in tokens

Sarah Friar argues that the familiar software metrics — seats purchased, active users, subscription renewals — fail to capture AI's real value.

"For years, software was measured through adoption: seats, active users, renewals.

AI is different: it must be measured by work completed," — Sarah Friar, Chief Financial Officer of OpenAI.

Friar adds that as their capabilities grow, models take on ever longer and more complex tasks: they hold context, reason across multiple steps, and work with several tools at once. The stakes are high: according to figures cited by CFO Dive, global spending on AI will reach $2.59 trillion in 2026 — 47% more than a year earlier.

"Tokens create value when they turn into work people can use,"

Friar writes on the OpenAI blog.

How many companies are earning back their AI investments

Only 12% of executives said AI had simultaneously cut their companies' costs and increased revenue, according to a January PwC survey cited by CFO Dive. Roughly another 33% of respondents saw an effect in at least one area — costs or revenue — while 56% of organizations have yet to record a meaningful financial result.

Friar's scorecard is a direct answer to that gap: OpenAI is urging finance chiefs to count not subscriptions and query volume, but the share of tasks brought to a useful result and their full cost. This approach makes it possible to compare AI projects against one another and against the "human" alternative.

What it means

The largest seller of AI is itself setting the framework for judging its payback — a signal that the market is moving from experiments to strict financial accounting. For companies, Friar's scorecard is a ready-made checklist: if the cost per successful task is not falling and "useful work" is not growing, it is time to rethink the AI budget.

Frequently asked questions

What is OpenAI's scorecard for the AI era?

It is a system for assessing AI payback built on four metrics — useful work, cost per successful task, reliability, and return on compute — which OpenAI CFO Sarah Friar presented on July 17, 2026 on the company's blog.

How should the cost per successful task be calculated?

Under Friar's methodology — as the full cost of obtaining a result: spending on AI usage, retries, and employees' time spent on review, not just the price of tokens.

Are companies' AI investments paying off today?

According to a January PwC survey, only 12% of CEOs see AI simultaneously growing revenue and cutting costs, while 56% of organizations have yet to obtain a meaningful financial effect.