AI adoption is up. Impact isn't.

Nearly nine in ten organisations (88%) say they use generative AI in at least one business function, up from 78% a year ago. This is the steepest single-year jump McKinsey has recorded in eight years of running this survey.  On the surface, this shows an industry finally moving past pilot purgatory. A closer look reveals a more nuanced picture.

Two-thirds of organisations use AI to power sevral functions. Agentic AI has leaped from novelty to mainstream (62% are trying it, 23% are scaling it somewhere), and high performers — the roughly 6% of firms actually redesigning workflows around the technology — are pulling ahead on innovation, customer satisfaction, and competitive differentiation.

What's overstated in most AI narratives is the leap from usage to value: Only 39% of respondents attribute any enterprise-level EBIT impact to AI at all. And most of that group estimates the effect at under 5%. Just a third of organisations have reached the scaling phase, meaning the majority using AI somewhere still haven't restructured a single workflow around it. Deployment has clearly outrun redesign — and redesign, not deployment, is what McKinsey's own data ties most strongly to measurable value.

"We use AI across the business" and "AI has moved the bottom line" are completely different claims. Before you cite adoption figures as evidence of progress, ask a better question: has a specific workflow changed as a consequence of using the tool, or has the tool simply been an added extra? If the answer is no, you're measuring enthusiasm, not impact. The good news is, that's a fixable problem, not a reason to slow down.

Forward this to team members if this resonates with you.