73% say they use AI. 10% say AI is core.
A survey of 1,550 leaders across six countries. The figure that caught me was not 73%, but the 42% who said AI is already good enough and their organisation cannot absorb it.
Publicis Sapient surveyed 1,550 AI decision-makers in the United States, the United Kingdom, France, Germany, Australia and the UAE, at businesses of 500 staff and $100 million in revenue or more. Fieldwork was late April into early May, published in mid-June.
Two figures sit beside each other in the report.
73% say AI is used regularly or across most processes. 10% say AI is core to how the business runs.
Sixty-three percentage points between those two sentences. That is where almost every company I work with is standing.
The figure that caught me was a different one
42% say AI is already capable enough, but their organisation is not built to absorb that value.
Set beside the 47% who believe AI can already meet today's business needs, the picture is fairly clear. Most of the people involved no longer doubt the tool. They doubt their own organisation.
Asked about the biggest barrier, 22% pointed at organisational operations. Not budget, not talent.
At small scale it shows more plainly
This survey covered large businesses. But the mechanism is the same, and a small company has nowhere to hide.
Last month I sat with a company of twenty people. They had bought AI accounts for the whole department back in March. I asked whether anything had been done differently in the past week.
A pause. Then one person said they used it to write captions, faster than before.
That was all of it. Six months, twenty accounts, one person writing captions faster.
Nobody is lying when they report that the company has adopted AI. It genuinely has. It is only that the way the company runs is exactly as it was.
The breakage I meet most often
Usually AI gets handed to “marketing, to try first”. It sounds sensible, and in practice it is handed to nobody, because nobody is accountable for reporting a result.
The second place is a goal stated as “use AI to be more efficient”. That is not a task. “Turn Monday's meeting notes into a to-do list before 10am” is a task, because it has an owner, a deadline, and the ability to fail.
And I have almost never seen anyone measure beforehand. If you do not know how many minutes the task currently costs, there is no way three months later to know what you gained.
If you are in the 73%, the question worth asking is not “have we adopted AI”. The question worth asking is: last week, did the company do anything differently because of AI?
Answer with one concrete example and you are in the 10%. Cannot answer, and you are somewhere in those sixty-three percentage points.
One more figure worth noticing
38% say AI is fundamentally changing how the business operates right now. Far higher than the 10% who call it core.
Two different questions, and the gap between them is where everything is happening. A great many organisations have genuinely changed, but not deeply enough to call it core.
That is the most uncomfortable position: money spent, effort spent, and not enough result yet to feel confident going further.
The way out is usually not buying another tool. Pick one process, finish it, measure it, and only then multiply.