Ask ten firms what their AI transformation looks like and most will describe a shopping list. A Copilot license for everyone. Two or three pilots running in different corners of the company. A vendor demo that looked impressive in the room. Then six months pass and nothing has changed in how the work gets done. The same tasks run the same way, with an AI step bolted on somewhere in the middle.
The numbers say this is the norm. In PwC's 29th CEO Survey, 56% of CEOs said AI hasn't delivered any significant financial benefit yet. MIT CISR found that only 7% of companies have reached the top stage they call "AI Future-Ready." Most firms sit well below it. The spend went out. The return didn't come back.
Here's the thing the shopping list misses. AI transformation is a shift in what your people can do that they couldn't do before. The tools are real, but they're the means. What changes a firm is the new capability and the new way of working that the capability makes possible. Buy the tools without that shift and you've bought a more expensive version of last year.
Why most transformations stall
You can't plan a route without knowing your starting point. That sounds obvious, and almost nobody does it.
Most firms jump straight to the tools because the tools are easy to name and easy to buy. The harder question is where the firm stands today. How are people already using AI? Who's steering it? Does any of it touch the work that matters? Without those answers, every plan is a guess, and most guess wrong in the same direction: too far ahead of where the firm really is.
This is why a maturity model helps. It gives you levels, so you can locate yourself before you plan a move. Most mid-market firms, when they look, land at the first or second level. People are experimenting on their own, leadership is starting to pay attention, and that's about it. There's nothing wrong with being there. The problem starts when a firm at level one buys a plan built for level four.
The five levels
This is the spine of Pacemark, the AI maturity model I built. Five levels, from scattered experiments to a firm that runs differently because of AI.
- Ad Hoc. People are using ChatGPT on their own. No one is steering.
- Pilot. One to three intentional experiments are running, and leadership is starting to pay attention.
- Integration. AI is embedded in core workflows. Policy and training exist.
- Acceleration. AI is a competitive edge. You've built proprietary workflows, and you track the ROI.
- Transformation. AI shapes how the firm wins work, delivers projects, and develops people.
Read those again and notice what moves between them. It isn't the number of tools. It's how deep the capability runs and how many people it reaches. A firm at level one and a firm at level four might own the same software. The difference is whether the work changed.
Most "AI transformation" talk skips levels two and three and sells the picture of level five. That's the trap. You don't get to transformation by buying it. You get there one level at a time, and the move from each level to the next is a different kind of work.
Transformation isn't one thing
A firm doesn't have one AI level. It has six.
Pacemark scores a firm across six dimensions: Strategy & Leadership, People & Culture, Data Readiness, Workflows & Operations, Technology Infrastructure, and Governance & Ethics. Each one gets its own level, one through five. Your headline Pacemark Index pulls them together into a single score.
Here's the part that matters most. The headline score isn't an average. It works like a bottleneck: your weakest dimension drags the whole thing down. A firm can sit at level four on Technology Infrastructure, with strong tools and clean integrations, and at level one on People & Culture, where almost no one trusts the tools or knows what to do with them. That firm is not a level four. The People score pulls it down, because the tools only return what people actually use.
Most maturity models bury the human side. They score the data, the tech, the infrastructure, and treat people as a footnote. Pacemark weights People & Culture equally with the technical dimensions, as its own dimension, scored on its own evidence. That's the difference, and it's the heart of the argument. Take your humans seriously or your transformation stays on paper.
Where transformations are won or lost
The human side decides it. Not the procurement, the platform choice, or the rollout plan.
I've watched this play out enough times to call it. A firm buys the tool, runs the training, hits a high completion number, and feels done. Six months later weekly use is low and no one can explain why. The tool was fine. The change never happened, because the change was always in the people, and no one owned that part.
This is its own discipline, and it separates the firms that get a return from the ones that don't. I wrote about it at length in AI change management. You can't install a capability shift. You lead one.
What this looks like in practice
Take a mid-market civil engineering firm. They thought they were already doing AI: two dozen Copilot seats, a board-approved policy, a few people producing real AI-assisted work. The Discovery scored them at Level 1. Half the seats sat unused, the policy had never reached employees, and the real constraint was strategy: no sponsor, no budget line, no one measuring anything. The capability was already in the building. Eleven people had built their own workflows on the side, one of them running document quality-control checks that cut weeks of review down to hours. Leadership knew about two of them. So the first moves cost almost nothing: name an executive sponsor, make those eleven people visible and give them real time, and point the firm at the work they were already proving out. The level moves when the work changes, and the people to change it were already there.
The shape of the story is always the same. The firm thinks it has a tool problem. The Discovery shows a bottleneck somewhere leadership didn't expect: strategy, the people, or how the work is organized. Rarely the tools. The first project targets that bottleneck instead of buying more software, and the level moves. Then the next move gets easier, because the firm can finally see what it's working with.
How to start: find your starting line
You don't need a strategy yet. You need to know where you stand.
The free Signal Scorer takes under ten minutes and gives you a Pacemark Index plus a radar across all six dimensions. It shows you your weakest dimension, the one setting your ceiling, before you spend a dollar on tools. Start there.
Once you can see the picture, the next question is the route. That's the work I do with founder-led firms: run the Discovery, score the firm, and build a plan sequenced so the first project pays for the next. If that's the help you're looking for, here's how I work as an AI strategy consultant.
The firms that get a return on AI build the new capability first. Then they buy the tools to run it. Transformation was never the software. It was always what your people can suddenly do.
