You've probably seen this number already. McKinsey's 2025 State of AI report: 88% of organizations have adopted AI in at least one business function. Only 6% are seeing meaningful impact on earnings. Most of them bought the platforms, trained the people, ran the pilots. No movement on the bottom line.
The stat gets cited constantly. What gets skipped is the more interesting finding underneath it.
The question that separates the 6%
McKinsey also looked at what distinguishes the high performers. Not which tools they bought, not how much they spent, not how early they started.
The 6% are three times more likely to have fundamentally redesigned their workflows.
That sounds abstract. Here's what it means in practice: the vast majority asked "how do we use AI to do our current work faster?" The 6% asked a different question. They asked "what does this work look like now that AI exists, and what should we charge for it?"
Same technology. Different question. Completely different financial outcome.
88% adopted AI. 6% see real earnings. The difference is the question they asked: not "how do we work faster?" but "what should we charge for?"
Take contract review. An AI drafting tool that produces the same contracts faster is 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's the other question. The first approach saves time. The second changes the business.
The ceiling on efficiency
Infosys surveyed 3,800 executives across 15 industries and ranked professional services #1 in AI viability. Only 15% of PS projects fail, compared to 26% cross-industry. Firms are getting real results from AI. But look at where: staff utilization, document review, compliance. Every use case Infosys measured is operational. Not one is about new revenue.
BDO's Nick Kervin put a number on the efficiency-only path: "Using AI to find efficiency in existing businesses will have a natural ceiling of 25-40%."
And even within that range, the gains are fragile. Brian Solis reported that nearly 40% of AI time savings are being lost to rework. Firms report faster delivery, then discover the output requires more human correction than the old process did. The ceiling is lower than it looks.
Most firms are already approaching it. The attorneys who've been using AI for a year are faster, but their 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.
Beyond that ceiling, the only path is changing the model. The framework view splits in two: The Pricing Shift covers what you charge, The Delivery Shift covers how the work shows up. That is not my opinion. It is arithmetic.
And the math looks different from the other side. Sequoia Capital's Julien Bek pointed out that for every dollar spent on software, six dollars get spent on services. The venture money that spent two decades chasing that software dollar is now chasing the six. If your firm is stuck at the efficiency ceiling, you're standing on exactly the ground those startups are built to take.
What the other question produces
When firms ask the redesign question instead of the efficiency question, the answers look different by vertical.
In law, firms 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. The firm's revenue per engagement goes up, not down.
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. The client pays a monthly retainer for visibility they didn't have before. That is new money, not discounted old money.
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. The revenue model shifts from project fees toward something more like a subscription.
None of these are hypothetical. Firms are already doing them. They are in the 6%.
What the clients are willing to pay for
Source Global Research asked tax clients directly: would you pay more for AI-enabled advisory if it generated more value? Every client surveyed said yes. Not a majority. All of them.
Clients aren't waiting for lower prices from AI. They're waiting for new value. If the vast majority of firms are still selling the same services at compressed margins, the firms that build genuinely new offerings aren't competing in a crowded field. They're selling something nobody else has.
One question to ask yourself
You've already adopted AI. Your team is faster. The question worth sitting with: is your AI investment building toward a different business, or just optimizing the one you have?
The 6% answered that question. Most firms haven't yet.