Employees report AI productivity gains. Most companies can't find them in operating profit.

In McKinsey's 2026 State of AI survey, 80 percent of respondents say AI has improved their individual productivity. Only 37 percent attribute any EBIT impact to AI, a share essentially unchanged from 2025. The gains are easier to report than to find in operating profit.

The survey's details matter. McKinsey fielded it online from May 4 to June 8, 2026, among 1,719 participants in 97 nations, and weighted the results by each nation's share of global GDP. Of the respondents, 36 percent work at organizations with revenue above $1 billion. EBIT, earnings before interest and taxes, is a standard measure of operating profit. Both figures are self-reported perceptions, not audited financials, and the EBIT measure counts any positive contribution, however small. The 37 percent is a share of respondents, not of companies or AI projects, so it should not be read as a failure rate.

The distance between the two numbers is where management comes in. Saved time reaches operating profit only when it turns into more or better output that customers pay for, or into lower costs. Neither happens by default.

Where the saved time goes

Employees report the time savings directly. BCG's 2026 AI at Work survey of 11,749 workers in 14 markets found that 42 percent of frontline employees who use AI regularly save at least a full workday a week. Yet 66 percent say they get limited or no guidance on what to do with that time, and over half do not redirect it to strategic work.

Two caveats apply. BCG defines frontline employees as individual white-collar employees without managerial responsibilities, not deskless or shop-floor workers. The time savings are self-estimated, and BCG's release does not state the base for the 66 percent, which appears to be regular frontline users.

If BCG's figures are close to right, much of the time AI saves arrives without instructions. Some of it may go into better work that no survey captures. Some may simply be absorbed. An organization that has not decided what the saved time is for has little reason to expect it to show up in EBIT.

What separates the high performers

McKinsey's survey does identify a group that reports the connection. About 6 percent of respondents qualified as AI high performers, meaning they attribute 5 percent or more of EBIT to AI and describe its value as significant. That share is unchanged from 2025. It is often misquoted as the share of companies getting any value from AI, when 37 percent report some EBIT impact.

The high performers stand out for redesigning work. Nearly three-quarters of them report fundamentally redesigning workflows because of their AI use, up from 55 percent in 2025, compared with one-quarter of other respondents. McKinsey also found them twice as likely to say senior leaders demonstrate commitment to AI and to have defined processes for measuring its impact. In their commentary, McKinsey's authors argue that adoption lags the technology because organizations struggle to absorb change.

BCG's results run in the same direction. Its 2026 survey associates a clear strategy with a 25 percentage point lift in measurable business impact from AI, against about 5 points for better tools alone. Respondents in companies pursuing workflow redesign were 24 points more likely to see measurable improvement and 22 points more likely to save a full day a week. The pattern fits the arithmetic of saved time. Redesign is the point at which an organization decides what the freed hours will do.

Neither finding is proof. Redesign correlates with higher reported impact, but companies already getting value may simply find redesign easier. The high-performer subgroup is small, so its percentages carry wide margins. BCG's lifts compare groups of respondents, not the results of controlled experiments, and its measure of impact is what employees perceive.

Saved time reaches operating profit only when it turns into more or better output that customers pay for, or into lower costs.

A measurement problem as well as a management one

The strongest objection to this argument is that the gap is partly an artifact of measurement. EBIT moves for many reasons at once, and isolating AI's share is hard. A company could gain real value from AI while survey respondents cannot point to it in EBIT. McKinsey's respondents also cite benefits that EBIT does not capture, such as innovation and employee satisfaction. BCG found that respondents in companies pursuing workflow redesign were 20 points more likely to report higher job satisfaction, a gain no income statement records.

A second objection is timing. In the same McKinsey survey, 44 percent report AI scaling across the enterprise, up from 38 percent. Benefits from that scaling may not have reached the accounts yet. The flat EBIT figure offers a partial answer: scaling rose over the year while the share reporting any EBIT impact stayed about the same. One year is too short to settle the question.

Both objections lead to the same practical step. If AI's value is hard to see in the accounts, measure it where it occurs, one workflow at a time, before scaling further.

1. Choose one workflow where AI use is already regular, and record its baseline cost and output.
2. Decide in writing what the saved hours are for, and tell the people doing the work. In BCG's survey, 66 percent said they get limited or no guidance on what to do with saved time.
3. Redesign the workflow around that decision instead of adding AI to the old process.
4. Track cost and output each month, and report the change in terms the finance team would accept as a contribution to EBIT.

A program that cannot show gains in a single workflow is unlikely to find them in operating profit.