McKinsey's 2025 State of AI report surveyed organizations across industries and asked whether AI was producing meaningful impact on earnings.
6% said yes.
Not 6% who were struggling. 6% of all organizations — including ones that have been investing heavily in AI for years, bought the platforms, trained the people, and run the pilots. 94% are using AI and not moving the needle on earnings.
What the 6% do differently
McKinsey also analyzed what distinguishes the high performers. Not which tools they use, or how much they spend, or how long they've been at it.
The AI high performers are three times more likely to have fundamentally redesigned their workflows — not just adopted tools.
The firms winning with AI didn't buy better tools. They changed what they do, not just how fast they do it.
Take contract review. Buying an AI drafting tool and using it to produce the same contracts faster is tool adoption. Asking "what does a contract review engagement look like now that AI handles the research and first draft, and how do we price and deliver that differently" — that is something else. The first approach saves time. The second changes the business.
The ceiling on efficiency alone
BDO's Nick Kervin put a number on this: "Using AI to find efficiency in existing businesses will have a natural ceiling of 25–40%."
Most firms are already approaching it. The attorneys who've been using AI for a year are faster, but their hourly billings are flat or declining because they're completing the same work in less time. The accounting firms that deployed AI tools in 2024 are processing more returns with the same headcount — which is fine until a client asks why the bill hasn't changed.
Efficiency within an existing model has diminishing returns. Beyond that ceiling, the only path is changing the model.
The opportunity in the number
There's something useful in the 6% finding, though. It's not just a warning — it's a market position.
If 94% of firms are using AI without meaningful financial impact, the firms that crack this aren't competing in a crowded field. They're in a different race.
Source Global Research asked tax clients directly: would you pay more for AI-enabled advisory if it generated more value? 100% said yes.
Not a majority. Every client surveyed.
Clients aren't waiting for lower prices from AI. They're waiting for new value. The firms that figure out how to deliver genuinely new services — not the same services faster, but services that only exist because of what AI makes possible — are positioned to charge more, not less.
What redesigning the workflow actually means
It looks different by vertical.
In law, the firms doing it are moving from hourly contract review to fixed-fee contract intelligence — AI handles the structural analysis and flags risk, attorneys apply judgment to the items that matter. Clients get a faster, clearer answer at a price they can budget for.
In accounting, the shift is from annual audits to continuous monitoring. AI watches the books in real time, the team surfaces findings when they're actionable rather than a year late, and the client pays a monthly retainer for visibility they didn't have before.
In consulting, it's the codification of expertise. AI delivers the analysis at scale, senior people focus on the decisions that require context and judgment, and the revenue model shifts from project fees toward something more like a subscription.
None of these are hypothetical. Firms are already doing them. The ones doing them are in the 6%.
The actual question
McKinsey's finding doesn't tell you whether your firm should adopt AI. You already have. The question is whether your AI investment is building toward a fundamentally different business model or just optimizing the one you have.
That's what separates the 6% from everyone else.