Three numbers from one survey, sitting two exhibits apart
The fourth annual BCG workplace AI study — 11,749 workers across 14 markets, June 2026 — carries three numbers most reports pull one at a time. Here they are together, because that is when they say something the individual figures do not:
42% of regular AI users say they save at least a full workday a week. 66% say they get limited or no guidance on what to do with the time. 47% now spend more time directing and managing AI than doing the underlying work.
The first figure is the one that lands in press releases. The other two describe what is happening to the first one after it leaves the announcement.
The productivity claim, made by people who were not told what the time was for
An hour freed is a real saving. If two thirds of the workers reporting it were also told nothing about where that hour was supposed to go, then a large share of the productivity number is being generated by workers with no assignment for the time. The organization booked the win at rollout. The accounting question — where the hour actually landed — is a separate one, and it has not been asked.
Where an unassigned hour goes
The 47% is what the answer looks like when the answer is left to the individual. An hour with no destination does not sit unspent — it goes back into the interface that just handed it over. One more prompt. One more revision. One more variation. The tool that produced the saving absorbs it, and the loop closes.
This is not a discipline failure at the desk. It is what predictably happens when a workflow change is announced without a workflow change being designed. The time is really there; nobody said what it was for.
A management job, not a technology one
The design work is smaller than it sounds. For each role that just gained an hour, name what that hour is for: a deeper review pass, a customer conversation the calendar never had room for, a piece of work being deferred because the ordinary week was full. The naming is the whole intervention. It moves the hour out of the tool it came from and into work that appears somewhere else you can measure.
The organizations that eventually see a P&L move from all this are the ones whose managers finished the second half of the change — the part that says where the hour goes — before booking the first half as a win. The ones running the same rollout on the assumption that individual workers will design the second half themselves are the ones whose productivity numbers stay on the announcement slide and never make it onto the finance report.
The hour has to arrive somewhere: Give the hour AI freed up somewhere to go →
Keep reading: Nobody decided where the hours go · The hour that arrives in pieces
This is the thinking behind a Trusted AI Culture.
Sources: Boston Consulting Group, "AI at Work: Why Strategy Matters More Than Tools" (fourth annual edition, June 2026), n=11,749 workers across 14 markets. Figures verified against BCG's 3 June 2026 press release. All figures are self-reported.
Written with AI in the loop: my idea, AI drafted and sharpened, my judgment on the way out. Every word is mine to stand behind. — Darren