AI maturity benchmarking has become a small industry. Consultancies sell five-level models, vendors give away assessments that conclude you need their product, and the reports land with charts showing you somewhere between “emerging” and “developing”. Most of it is built for companies with a Chief Data Officer and a data platform team.
If you run a business with 5 to 50 people, you need something simpler. This playbook gives you a five-question check you can do in 15 minutes, explains what maturity looks like at your scale, and shows how to move up one level without hiring anyone.
Why most AI maturity models are useless for SMEs
The big maturity frameworks were designed for large organisations, and it shows. They score you on things like enterprise data architecture, a centre of excellence, an AI ethics board and a portfolio of models in production. A 12-person accountancy practice will score badly on every one of those and learn nothing useful from the result.
They also measure the wrong thing. Most models reward ambition and infrastructure. What a small business needs to know is much plainer: are we getting value from AI, safely, and is that value growing?
There’s a third problem. Self-assessments flatter. Ask a management team whether AI is “embedded in key processes” and you’ll get a confident yes from people who mean “two of us use ChatGPT most days”.
The five-question maturity check
Answer each question honestly and score it 0, 1 or 2.
1. Do you know who’s using AI and for what?
- 0: No idea.
- 1: Roughly, from conversation.
- 2: Yes, it’s written down in a register, including the tools and accounts.
2. Do people know the rules?
- 0: There are no rules.
- 1: There are rules, but they live in someone’s head or an email from last year.
- 2: There’s a short written policy, and staff have read it.
3. Has AI changed at least one process, with a measured result?
- 0: No.
- 1: People say it saves time, but nobody has measured it.
- 2: Yes, and you can name the process and the before-and-after figure.
4. Can AI safely reach the information it needs?
- 0: Files are scattered across email, desktops and personal drives.
- 1: Mostly in one place, but permissions are a mess.
- 2: Organised in a business system, with sensible access controls, and connected to a business-grade AI tool.
5. Has your team been trained beyond a demo?
- 0: No training.
- 1: A webinar or a lunch-and-learn.
- 2: Practical training on their own tasks, with follow-up.
Add up your score out of 10.
| Score | Level | What it usually means |
|---|---|---|
| 0 to 3 | Experimenting | Individuals are trying things, mostly on personal accounts. Value is real but invisible, and so is the risk. |
| 4 to 6 | Using | AI is part of the working week for some people. There’s some structure, but results depend on who you ask. |
| 7 to 8 | Embedding | AI has changed specific processes, the rules are clear and you can point to measured gains. |
| 9 to 10 | Compounding | Gains are measured, shared and reinvested. Improvements in one area spread to others. |
If you want a fuller picture, you can request our free AI Readiness Scorecard for a review across the five Trust Stack layers. The five-question check above is a useful conversation starter for a management meeting.
What real maturity looks like
Real maturity in a small business is boring. That’s a compliment.
Picture a solicitor’s practice where every standard letter starts as a draft in an approved AI tool and gets checked by a named person before it goes out. The practice manager knows the process saves about 4 hours a week, because she timed it before and after. Maturity also looks like a quarterly 15-minute review where someone asks which tools are still earning their keep, and an induction pack with a page on AI.
What it doesn’t need is a strategy document. The businesses I’d call mature at this scale tend to have fewer AI tools than their neighbours, used more consistently, with someone clearly responsible.
The maturity gap that actually matters
The national figures show how wide the spread is. The CSO’s 2025 enterprise survey found that 20.2% of Irish enterprises used AI technologies, but only 17.2% of small enterprises did, compared with 57.7% of large ones. Ireland is level with the EU average of 20%, and well behind Denmark at 42%.
Usage is only the first gap, though. The more telling one is between using AI and getting value from it. The AI Economy Ireland 2026 report from Trinity College Dublin and Microsoft Ireland found that 92% of organisations use or plan to use AI, but only 10% of leaders describe their deployment as advanced. Large organisations were more than twice as likely as SMEs to save 2 or more hours per employee per week (54% against 25%).
And here’s the figure I’d pin to the wall. Organisations with a formal AI policy were 10 times more likely to report major productivity gains: 30% against 3%. That’s a correlation, so it doesn’t prove that writing a policy causes the gain. My reading is that a policy is a sign of a business that has made deliberate decisions about AI, and deliberate decisions are what separate the levels in the table above.
There’s one more gap worth checking inside your own walls: the distance between your best AI user and your typical one. If one person is getting 5 hours a week back and everyone else is getting nothing, your maturity is a single point of failure.
How to move up one level
Don’t try to jump two levels. Pick the move that fits where you are.
From experimenting to using. Run an amnesty and find out what people already use. Put it in a register. Choose one approved tool and get everyone onto a business account. Our shadow AI article has a register template you can copy.
From using to embedding. Pick one process that eats time every week, such as proposals, reporting or customer emails. Time it now. Redesign it with AI, train the people involved, and time it again after a month. Write down the result. That’s your first piece of evidence, and the evidence layer of the Trust Stack explains why it matters more than any strategy.
From embedding to compounding. Share what’s working. Get the person who made the first process work to show the rest of the team, then repeat the exercise on the next process. Review your tools and policy each quarter.
Whichever level you’re at, the next step is smaller than the maturity industry would have you believe.
If you’d like help working out where you stand and what to do next, our AI readiness and planning service starts with exactly this conversation, and the first 30 minutes are free.