AI readiness assessment: 12 questions to score your business
Easy Insight Team ·
AI readiness is mostly not about technology. It is about whether your data is findable, your processes are written down, and someone is accountable for what the system does. These twelve questions score all three in about an hour, and produce a number that tells you whether to start, pilot, or fix the foundations first.
Why readiness rather than strategy?
Because the national picture says the bottleneck is not ambition.
Around 35% of UK businesses with 10 or more employees reported using at least one AI technology in June 2026, up from roughly 12% in late 2023. But the average adopter uses only about 1.6 AI technologies, up from 1.4, and only around 10% report extensive use (ONS, published July 2026).
That is a picture of adoption spreading sideways rather than deepening — lots of businesses with a tool switched on, very few that have changed how work happens. The gap between those two states is what readiness measures.
The twelve questions
Score each 0 (no), 1 (partly), or 2 (yes). Answer honestly; a flattering score helps nobody.
Data
1. Can you get a list of your customers, orders or cases without asking a specific person? If the answer is "ask Sharon, she has the spreadsheet", score 0. Single points of human failure are the most common blocker we see.
2. Do your key numbers agree across systems? If the CRM, the accounts package and the board pack give three different revenue figures, AI will confidently pick one.
3. Is your data stored somewhere a system could reach it? Files in a shared drive count. Files in individual inboxes, or in a filing cabinet, do not.
4. Do you keep enough history to see a pattern? Twelve months of consistent records is a reasonable floor for anything predictive. Three months of records plus a system migration is not.
Process
5. Could you write down the process you want to improve, in one page, today? If two people would write different pages, the process is not ready to automate — automating it would simply broadcast the disagreement faster.
6. Do you know how long it currently takes and what it currently costs? Without a baseline you cannot prove the project worked, which means you will argue about it for a year.
7. Is the process high-volume or high-value? AI pays back on things you do hundreds of times, or on things where a single error is expensive. A task done twice a month is rarely worth it.
8. Does the process have a clear right answer? Classifying invoices has a right answer. "Improving customer relationships" does not. Start with the former.
People and governance
9. Is there a named person who will own this after the consultants leave? Not a committee. A person, with time allocated.
10. Have you settled the data protection position? If the system touches personal data, UK GDPR applies. The ICO's guidance on AI and data protection states that in the vast majority of cases the use of AI involves processing likely to result in high risk to individuals' rights and freedoms, which triggers the legal requirement to carry out a DPIA.
11. Do your staff already use AI tools, and do you know which? If people are pasting client data into consumer chatbots, you have an AI programme already — an ungoverned one. Knowing scores 2; suspecting scores 1.
12. Can you say what "better" looks like as a number? "Cut average handling time from 11 minutes to 6" is a target. "Be more efficient" is a wish.
How to read your score
Out of a possible 24:
| Score | What it means | What to do next |
|---|---|---|
| 0–12 | Foundations first | Fix data access and process documentation. An AI project now would fail for reasons that have nothing to do with AI. |
| 13–18 | Pilot territory | Pick your single best-documented process and run one narrow proof of concept with a pass/fail test. |
| 19–23 | Ready | Start with a production build on one process, and use it to build the pattern for the next. |
| 24 | Suspiciously good | Re-score with someone who disagrees with you. |
The distribution matters as much as the total. A business scoring 20 with a zero on question 9 — nobody owns it — is not ready, whatever the total says. Ownership and data access are the two that cannot be compensated for elsewhere.
What if the score is low?
Low readiness is a reason not to buy a large AI project. It is not a reason to do nothing.
Almost every business that scores badly still has one process that is well understood, high volume and documented — and starting there is precisely how readiness improves. You learn what your data is really like, you find out who will own things, and you get a result small enough to argue about cheaply.
What low readiness should stop is the six-figure programme built on data nobody trusts. That is the pattern behind most stalled AI projects: the technology worked and the inputs did not. If questions 1 to 4 scored badly, the honest next move is data work rather than AI work — the same argument as recovering from spreadsheet chaos, and why a data strategy usually has to come first.
If you would like the score pressure-tested by someone with no incentive to tell you you are ready, that is what our AI strategy and readiness work is for, and the wider AI practice page sets out what follows it. What a readiness engagement costs is covered in our AI consulting cost guide.
Frequently asked questions
What is an AI readiness assessment?
A structured check of whether your business can actually get value from AI yet — covering your data, your processes, your people and your governance. It is deliberately not a technology review. The questions that decide whether AI works are almost all about how organised the business is, not which model you pick.
What score do we need before starting?
On the twelve-question scale above, 19 or above means you are ready to build on one process. Between 13 and 18, run one narrow proof of concept while fixing the obvious gaps. Below 13, the honest answer is that groundwork will produce a better return than an AI project this year.
Is low readiness a reason not to do anything?
No. It is a reason not to buy a large AI project. Almost every business scoring low has one well-documented process that would benefit from automation, and starting there is how readiness improves. What low readiness should stop is a six-figure programme built on data nobody trusts.
Do we need a DPIA before we start?
Often, yes. The ICO's guidance on AI and data protection states that in the vast majority of cases the use of AI involves processing likely to result in a high risk to individuals' rights and freedoms, which triggers the legal requirement to carry out a data protection impact assessment. Question 10 is there for exactly this reason.
How long should an assessment take?
The questions take an hour with the right three or four people in the room. Gathering honest answers to questions two and three — where your data actually lives and whether people trust it — is what takes longer, and the difficulty of getting those answers is itself part of the result.
Easy Insight is a UK consultancy for AI, web, apps and data — senior specialists only, no juniors.
Next step
Wondering where AI would actually pay in your business?
An AI strategy and readiness review (from £2,500) tells you where AI pays, where it doesn't, and what to leave alone. The price is fixed in writing before we start, and the advice comes from a team that runs AI in its own products.