Free tool · AI
AI readiness scorecard
Answer for one specific task you want handled, not for the business in general. A vague question produces a useless score.
12 questions across 5 dimensions — use case, process, data, ownership and commitment. It returns a weighted score and names your blockers. For most businesses the honest answer is not yet, and the tool says so rather than selling you a project.
The scorecard
Twelve questions, answered honestly
Pick the option that is true today, not the one you intend to be true by the time anything is built.
Answered and scored entirely in your browser. Nothing is submitted, stored or sent, and there is no email step — the result is the result.
Method
What this is actually testing
Five dimensions, weighted unequally. Process maturity carries the most weight, then use case clarity and data readiness, then ownership, and budget least of all. That ordering is the opposite of how these conversations usually start, which tends to begin and end with what it costs.
The reason is that failed AI projects fail in a recognisable pattern. The pilot impresses everybody, then it meets the cases nobody documented, then the person who championed it moves on, and eighteen months later it is quietly switched off. None of those failures are technical, and none of them are budget. They are process, data and ownership — which is why enthusiasm cannot rescue a zero in those columns here.
The weights are our judgement from delivery work, not published research, and are stated as judgement. What they encode is a bias towards telling you to wait.
A low score is a useful result, not a failing grade. The blockers it returns are nearly all things you can fix internally in a few weeks — and in a fair number of cases, doing that tidying delivers most of the saving on its own and no AI project turns out to be needed at all.
Questions
About this scorecard
Why does it weight process higher than budget?
Because a well-funded project on an undocumented process fails more reliably than a thin budget on a clear one. Automation encodes a process — if the process only exists in somebody’s head, the build turns into an argument about what the process actually is, conducted at consultancy rates. Money does not resolve that argument; writing it down does.
I scored badly. Does that mean AI is not for my business?
No, it means not yet and it names why. The blockers it returns are almost always fixable in weeks, internally, by people who already know the business. What a low score genuinely rules out is commissioning a build right now — that would mean paying a supplier to discover things you could write down yourself for nothing.
What does a realistic score look like?
Most businesses asking about AI for the first time land between 30 and 55. Scoring above 80 is unusual and normally means you have already run a project and learned from it. If you scored very high on a first attempt, re-read the process and data questions — those are the two people most often answer aspirationally.
How is this different from your AI readiness assessment service?
This is twelve questions you answer about yourself in five minutes. The paid assessment involves us in your systems, interviewing the people doing the work, and quantifying the actual time and cost of the task. If this scorecard tells you to go away and document a process, do that rather than buying the assessment — the assessment is worth having once you are past that.
Does it cover the legal side of putting data into a model?
Only as one question, and deliberately lightly. It asks whether you have checked that you may lawfully put the data through a third-party model, because an unchecked answer there is a genuine blocker. It is not a DPIA and it is not legal advice — if you are handling personal or special category data, that assessment is a separate piece of work.
Can I answer this for a specific project rather than the business overall?
Yes, and you should. The questions are framed around one named task on purpose. Answering them for "the business" in general produces a vague score that tells you nothing — pick the single task you would most like handled and answer about that.
Related reading
The other free tools
Scored low and not sure what to fix first?
Twenty minutes on a call will usually settle it — and if the answer is that you should spend a month tidying before spending anything with us, that is what we will say.