The AI readiness assessment for UK SMEs

Most business owners I speak to want to know one thing before they commit: is my business ready for AI? It is a sensible question. Nobody wants to spend money on a project that fails because the basics were not in place.

The problem is that the usual answer is a shrug and a gut feel. Someone reads a few articles, talks to a vendor, and decides yes or no based on whether the vendor seemed trustworthy. That is a coin flip with extra steps.

A proper AI readiness assessment is the answer. It scores your business across the dimensions that actually predict whether an AI project will work, gives you a number you can compare against a benchmark, and tells you exactly what to fix if you are not quite there yet. After running this assessment with dozens of UK SMEs, here is the framework we use, so you can self-assess in about ninety minutes.

What an AI readiness assessment actually is

An AI readiness assessment is a structured scoring exercise across the seven dimensions that determine whether an AI project will deliver value or quietly die in month three. It is not a personality quiz. Done properly, it gives you an honest read on where you sit today, which gaps to close first, and whether to pilot now or wait a quarter.

It works because AI project failure is almost never about the AI. It is about prerequisites. What fails is the data nobody could find, the process nobody had written down, the champion who got pulled onto another project, or the success metric that turned out to mean different things to different people. An AI readiness assessment surfaces these before you spend money.

Here are the seven dimensions we score, what each one means, and what to look for.

Dimension 1: Data hygiene

Data hygiene comes down to three things. Do you have the data, can you find it, and is it consistent enough to trust.

You do not need a data warehouse. You need to know where your customer records live, whether your order data is roughly accurate, and whether someone could pull a list of last quarter's invoices without a two-day expedition. Most SMEs score three out of five here. A pilot can run on imperfect data, and the AI often helps clean things up as a side effect.

What kills projects is not knowing what you have. If you cannot answer "where does this data live and who owns it" in under a minute, score a two and put a name against each dataset before you do anything else. The ICO guidance on AI and data protection is worth a skim, because data hygiene and GDPR exposure are closely linked.

Dimension 2: Process maturity

Process maturity is whether the work you want AI to help with is repeatable. If two people doing the same task would produce broadly the same output, the process is mature enough. If every job is bespoke and the steps live entirely in one person's head, it is not.

The test I use is simple. Could you write a one-page set of instructions that would cover eighty percent of cases? If yes, score yourself a four or five. If you could write the instructions but they would only cover half the cases, score a three. If nobody has ever written it down because it is "just what we do", score a two.

This is the dimension most owners underestimate. Processes feel obvious from the inside, but a morning spent shadowing the person who actually does the work usually reveals fifteen undocumented decisions per hour. You need to know about them before you try to automate them. See our piece on where to start with AI if you run a small business for how to pick the right first process.

Dimension 3: Team capability

Team capability is digital literacy plus openness to change. Not whether your team can code. Whether they can use a spreadsheet without panic and whether they treat new tools as opportunities or threats.

Most UK SME teams score better here than their owners expect. If your team uses email, a CRM, an accounts package, and a messaging app without daily complaints, they have enough digital literacy. The interfaces are designed to look like the things they already use.

The openness side is where it gets interesting. AI works best when the people doing the work help shape it. If your team has been burned by previous tech rollouts where nobody asked them anything, expect resistance and budget time to involve them properly. A four here with a willing team beats a five with a resentful one.

Dimension 4: Tooling baseline

Tooling baseline is about what your systems can talk to. AI projects almost always involve reading data from one system and writing results into another, so the question is whether your existing tools have APIs, integrations, or at least exports.

Cloud-hosted tools generally score well. On-premise software from the last decade usually has integration options. What scores low is anything sitting on paper or trapped in a PDF that someone retypes every morning. Not a dealbreaker, but a flag.

If your business runs on a mix of modern SaaS tools, score a four or five. If you have one critical system nobody has been able to integrate with for years, score a three and plan for it. If the core of your operation is paper or spreadsheets emailed around, score a two and start there. None of this rules out AI, it just changes the starting point.

Dimension 5: Leadership engagement

Leadership engagement is the single strongest predictor of AI project success in smaller organisations, according to MIT Sloan Management Review research. Not budget. Not data quality. Leadership engagement.

In practice that means one named champion with three things: authority to make decisions without convening a committee, curiosity about whether the project will work, and a few hours a week of genuine availability. Enough to review outputs, give feedback, and push when the project hits the inevitable wobble in week three.

Score a five if you have a named champion actively asking when the pilot starts. Score a three if leadership is supportive but distracted. Score a two if AI is something everyone agrees is important but nobody has put their name against. A low score here is the most common reason promising projects die quietly, and the easiest to fix.

Dimension 6: ROI clarity

ROI clarity is whether you can describe what better looks like in numbers. "More efficient" is not a number. "Reduce order processing from twenty minutes to five minutes" is.

This sounds obvious and catches almost everyone out. When I push business owners on what specifically they want AI to improve, the first answer is vague, the second more specific, and by the third pass we have something measurable. That third-pass answer is what the assessment is looking for.

You need a baseline, a target, and a rough estimate of what hitting that target is worth in pounds or hours. If you have all three for at least one process, score a four or five. If you have a target but no baseline, score a three and spend a week measuring. Our post on why most AI projects fail before they start goes deeper.

Dimension 7: Risk posture

Risk posture is about GDPR, customer data exposure, and any regulatory rules your sector layers on top. It is not about avoiding risk. It is about knowing what risk you are taking on.

For most SMEs the questions are straightforward. Will the AI process personal data, and if so, do you have a lawful basis. Where will the data be processed and stored. Who in your business signs off on data protection decisions. If you can answer these without a long pause, score a four or five. If you are not sure, score a three and book an hour to get clear before the project starts.

Regulated sectors (financial services, healthcare, legal) need to layer their own regulator's guidance on top. The good news is that the questions are familiar to anyone in those sectors already. The bad news is that pretending the questions do not exist is exactly how a project gets stopped six weeks in. Better to address it upfront.

How to self-score

For each dimension, give yourself a score from one to five.

• 1 means this is genuinely not in place and would block a project today
• 2 means it exists in patches but is unreliable
• 3 means it is broadly there with known gaps
• 4 means it is in good shape with minor things to tidy
• 5 means it is a genuine strength

Be honest. The point is to find out what to fix, not to score well. Optimistic scoring is the most common way businesses talk themselves into a project that then disappoints.

What your total score means

Add the seven scores together for a number out of 35.

• Under 21: Not yet. You have two or three significant gaps that would scupper a pilot. Fix those first. Usually it is leadership, process documentation, and ROI clarity in some combination.
• 21 to 28: Pilot-ready. You have enough in place to run a focused pilot on one process, learn from it, and decide whether to scale. This is where most UK SMEs sit.
• 28 and above: Ready to scale. You can confidently run multiple AI projects in parallel and start thinking about a broader programme rather than individual pilots.

These bands are deliberately forgiving because the dimensions are weighted equally in this self-scoring version. In our internal assessments we weight leadership and ROI clarity more heavily, because those are the two that most predict outcomes. If both of those are a two, treat the total score with caution regardless of what the others say.

Common failure modes the assessment catches

Three patterns come up repeatedly. A five on tooling and a two on leadership: the tech is ready, nobody owns it, it dies. A five on ROI clarity and a two on process maturity: the target is clear, the underlying work is too inconsistent to automate, results vary wildly. Strong scores everywhere except risk posture, where a GDPR question surfaces in week six and the whole thing gets paused.

Catching any one of these before you start saves the cost of the pilot and the morale hit of a failed project. For a deeper look at the same patterns, see our three signs your business is ready for AI.

What to do next

If you want a done-for-you version of this AI readiness assessment, that is exactly what our free AI opportunity report is. You tell us about your business, we run the seven-dimension scoring, and we send back a written report with your scores, the two or three things to fix first, and a shortlist of pilot opportunities with rough cost and return estimates. We turn it around within two business days.

We are also shipping an interactive AI readiness checklist tool shortly at /tools/ai-readiness-checklist, which will let you self-score in the browser and download a PDF summary.

Get your free AI opportunity report here and find out exactly where your business sits on the AI readiness assessment.