Enterprises are writing seven and eight figure cheques for AI. Licences, platforms, custom builds, consulting engagements to bolt it all together. Nobody's blinking at the invoice anymore.

What almost nobody is doing is asking the follow-up question: is anyone actually using it?

Not "have we rolled it out." Not "is it in the environment." Used. Daily, in the workflow, changing how someone does their job. That's a different bar, and I don't think most organisations have any idea whether they're clearing it.

The Measurement Gap

Ask a CIO how much they've spent on AI this year and you'll get a number to the nearest dollar. Ask what percentage of the workforce has changed a single habit because of it, and you'll get a shrug, a login-count dashboard, or a story about one enthusiastic team in finance.

Login counts and licence utilisation aren't adoption metrics. They tell you the tool was opened, not that it changed an outcome. Real adoption shows up in things like cycle time, error rate, or output per head — metrics that live outside the AI vendor's dashboard entirely, which is exactly why they're so rarely tracked. If the only measurement on the table is seats provisioned, the business has confused deployment with adoption. That confusion is expensive.

Attitude or Capability?

When adoption stalls, the instinct is to call it resistance. Sometimes it is. But resistance is a symptom with at least two very different causes, and treating them the same way wastes another budget cycle.

Capability gaps look like resistance but aren't. Someone doesn't know how to write a decent prompt, doesn't trust the output because nobody's shown them how to verify it, or hasn't had ninety uninterrupted minutes to sit with the tool. That's solved with training and time, not incentives.

Attitude gaps are a different problem entirely. Here the person understands the tool perfectly well and has decided not to use it — because it threatens how they're evaluated, because leadership rolled it out as theatre rather than strategy, or because the last three transformation initiatives went nowhere and this one has the same smell. No amount of training fixes that. It's a trust and incentive problem, and it usually starts above the person you're looking at.

Confusing the two is how a company spends a year running lunch-and-learns to fix what was actually a leadership credibility problem.

Top-Down or Bottom-Up?

The direction of resistance matters as much as its cause.

Top-down resistance looks like a mandate everyone nods at in the town hall and ignores by Wednesday, because middle management was never given a reason to change how they run their team — only an instruction to.

Bottom-up resistance looks like a genuinely useful tool that individual contributors love and quietly use for their own work, but that never makes it into a single official process, because nobody above them owns making it the standard way of doing things.

Both produce the same surface symptom — low measured adoption — and the same wrong diagnosis gets reached for constantly: mandate harder at bottom-up resistance, or throw more grassroots champions at top-down resistance. Neither works. You have to know which one you've actually got before you spend a dollar fixing it.

Why Was the Money Spent in the First Place?

This is the question that should have been asked before procurement, and almost never was.

Was the investment made to solve a named, costed business problem — too many hours lost to manual reconciliation, too much attrition on a specific process? Or was it made because a board member asked what the AI strategy was, and "we bought some licences" was the fastest defensible answer?

Spend justified by a real problem has a natural adoption owner: whoever owned that problem before. Spend justified by competitive anxiety usually doesn't. It lands in IT, gets provisioned, and waits for someone to find a use for it. That's not really an adoption failure — it's an adoption plan that was never written, because the purchase was never about adoption to begin with.

So Who Owns It?

Ask most enterprises who's accountable for AI adoption and you'll get one of three answers: IT (who own the licence, not the workflow), a transformation office (who own the programme, not the outcome), or nobody in particular. All three are the wrong owner, because none of them sit close enough to the work to know if it's actually changed.

Adoption accountability belongs with whoever owns the P&L or the operational metric the tool was meant to move. If it was bought to cut proposal turnaround, the head of sales ops owns adoption — not IT. If nobody can name that person, that's the real finding. Not "adoption is slow," but "nobody was ever on the hook for it."

Where the Partner Ecosystem Actually Earns Its Keep

This is where it gets relevant to anyone selling through AWS, or GCP, or Microsoft, or the AI labs themselves.

Right now there's a stampede of technical partners all telling the same story to the same hyperscalers: we can build sophisticated AI solutions for your customers, we can drive consumption, put us in front of the account team. It's a crowded, noisy pitch, and most of it sounds identical because it's all optimised for the same metric — usage going up, workloads getting deployed.

Almost nobody is standing up and saying: we're the partner who makes sure this actually gets adopted business-wide. That's a different specialism entirely, and it's sitting there almost unclaimed. A partner who can walk into AWS, GCP, Microsoft, or Anthropic and OpenAI directly and say "we don't just build the thing, we make sure the workforce actually changes how it works because of it" isn't competing in the consumption stampede at all. They're solving the problem the hyperscaler needs solved to keep the consumption story alive past year one, which is exactly the problem very few in the market are capable of solving.

That's the opportunity. Not "bring adoption expertise to a co-sell deal," but recognise that adoption is the one gap in the entire partner ecosystem that isn't already saturated. Every hyperscaler has hundreds of partners telling them they can drive deployment. Vanishingly few can tell them they can drive genuine, sustained adoption after the deployment is done, or even drive the change management needed to take on the project in the first place.

If you're a partner sitting on real change-management or enablement capability, this is the moment to lead with it. Not as a nice-to-have bolted onto the technical pitch, but as the headline. The consumption story is table stakes now. The adoption story is still wide open.

One to Read: Worth searching out Gartner's ongoing research into the gap between AI investment and workforce trust in outputs — it puts real data behind most of the anecdotes above.

Thanks for reading. If your org has spent big on AI and genuinely doesn't know whether it's landed, that's exactly the kind of conversation I enjoy having — feel free to reach out.