What is an AI readiness assessment, and what does it measure?

An AI readiness assessment is a structured set of questions that scores how prepared your business is to get value out of AI, across six areas: Data, Process, People, Technology, Strategy and Governance. It is a diagnosis, not a product recommendation. You answer questions about the state of your business, you get a score, and the score tells you which of those six areas will sink an AI project if you spend money before fixing it.

The free version of ours is 30 questions, five per pillar, and takes 7 minutes. At the end you get a readiness score out of 120, a band for each of the six pillars, and a short synopsis. That is the whole free tier. If you want the action plans you pay £97 for the Full Report, and if you want a 30/60/90-day plan for closing the gaps you pay £497 for the Deep Dive. No price on application. No call required, no demo, no sales contact unless you ask for one.

Why an assessment exists at all

RAND’s analysis of enterprise AI projects found 80.3% fail to deliver the business value promised. The breakdown matters more than the headline: 33.8% are abandoned before they ever reach production, 28.4% reach production and underdeliver, 18.1% run but never recover the investment, and 19.7% actually deliver on the business case (AI Assembly Lines).

Read that split again. Two thirds of the failure happens before the model is the problem. Projects die because the data going in was wrong, or nobody documented the process the AI was supposed to accelerate, or the team quietly went back to the spreadsheet, or no single person was accountable for the outcome. Those are conditions inside your business, and they are all knowable in advance. An assessment is simply the act of finding out before you sign a contract rather than nine months into one.

The pattern repeats at board level too. PwC’s Global CEO Survey of 4,454 CEOs across 95 countries found 56% report neither revenue gains nor cost reductions from their AI initiatives, and only 12% got both (PwC). That is not a technology failure. That is a readiness failure, discovered late and expensively.

The six pillars, and what each one is actually testing

Five questions per pillar. Thirty questions total. Here is what each set is looking for.

Data

The opening question is deliberately uncomfortable: if you needed a clean, accurate list of your best 500 customers right now, how confident are you in what you would get? Most people answer honestly and then go quiet for a moment.

The rest of the pillar asks how complete the data behind your decisions is, whether your key systems can share data without someone manually exporting and importing spreadsheets, how often reliable data actually drives an important decision rather than instinct dressed up as experience, and who owns data quality by name.

That last one is the one that predicts failure. If the answer is “everyone” or “IT, I think”, the AI project will inherit the same ambiguity. We have written separately on why dirty data is the cheapest AI mistake to fix first, because it genuinely is.

Process

AI applied to an undocumented process automates the version of it that happens to be in front of you that week. So the Process pillar asks how many of your core processes are documented well enough that a new starter could follow them without asking anyone. It asks what happens if three different people in your business are given the same task: would their approach look similar, or would you get three different answers and three different outputs?

Then it asks how often you review and improve those processes rather than just running them, whether you know where the bottlenecks are, and again, who is responsible for making sure processes are followed and improved.

People

This pillar is about capability and behaviour, not enthusiasm. It asks you to describe your team’s attitude to AI honestly, which is a different question from describing it on a slide. It asks how capable they would be if you told them to use an AI tool in their day-to-day work tomorrow. It asks how actively you, as the leader, are modelling adoption: trying it yourself, talking about it, normalising it. And it asks whether the business has made any formal investment in helping the team understand it.

Leadership modelling is in there for a reason. A tool nobody senior uses becomes a tool nobody uses.

Technology, Strategy, Governance

The final three pillars cover the platform you would be building on, whether there is a defined commercial outcome attached to any of this, and whether anyone owns the consequences.

Governance is where the scores tend to collapse. Grant Thornton’s 2026 AI Impact Survey found 78% of business executives lack strong confidence they could pass an independent AI governance audit within 90 days (Grant Thornton). Schellman’s 2026 State of AI Governance Report shows why: 42% of organisations put AI purchasing authority with the CIO or head of IT, 16% with a Chief Data Officer or AI Officer, and only 10% with the CEO, which means in most businesses the executive who buys the tool is the same executive blamed when it creates a compliance problem (Schellman). That is a single point of failure, not a governance structure. If you want the longer argument, see who owns AI outcomes in a company.

What a score out of 120 actually means

Thirty questions, four points available on each, 120 at the top. The total is the least interesting number on the page.

What matters is the shape. A business scoring 78 with even bands across all six pillars is in a very different position from a business scoring 78 with strong Technology and Strategy and a floor-level Data band. The first one has a general lift to do. The second one has a specific, expensive problem: it has bought platform capability it cannot feed. That business will run a pilot, the pilot will work on a hand-picked dataset, and the rollout will fail on the real one.

So read the pillar bands first and the total second. The lowest band is your constraint. Money spent anywhere other than the lowest band is money spent on a bottleneck you have not moved.

This is also why a working pilot is weak evidence. A survey of 650 enterprise technology leaders in March 2026 found 78% of enterprises have AI agent pilots running but fewer than 15% reach production (Digital Applied). Pilots are run on the best data, by the most willing people, with governance suspended. An assessment measures the conditions a rollout will actually meet.

What the assessment does not do

I would rather be straight about the limits than oversell the instrument.

It does not audit your systems, your data or your code. Every answer is self-reported by the person taking it. If you answer optimistically, you get an optimistic score and a useless one. It does not review the AI tools you have already bought or check whether anyone still uses them. It is not a call, a consultation or a demo. And the free tier does not give you an action plan. It gives you a score, six bands and a synopsis, which is enough to know where you stand and not enough to fix it.

That last distinction is the honest reason the paid tiers exist. The £97 Full Report is the same diagnosis with action plans attached. The £497 Deep Dive turns it into a 30/60/90-day plan for closing the gaps before you spend on AI. Same diagnosis, different depth of course correction.

AI readiness assessment versus AI maturity model

These get used interchangeably and they are not the same thing.

A maturity model describes how far along a curve you are, usually in five stages with names like “emerging” and “optimised”. It is descriptive and comparative. It tells you where you sit relative to a model.

A readiness assessment is predictive and specific. It asks whether the conditions for a particular kind of spend to succeed are present in your business right now. You can be at a low maturity stage and perfectly ready for one narrow use case, because the data that use case needs happens to be clean. You can be at a high maturity stage and unready, because nobody owns the outcome. Maturity tells you a story about your organisation. Readiness tells you whether to write the cheque this quarter. For the fuller version of this argument, see what does AI readiness actually mean for a mid-sized business.

Who should take it, and who should not

The assessment assumes the person answering can see across the business. A CEO, MD, COO, CFO, CIO, CTO or IT Director can answer all 30 questions with reasonable accuracy. A functional head can answer their pillar well and guess at the other five, which produces a score that looks precise and is not.

The better move in a 500 to 5,000-person business is to have two or three leaders take it independently and compare. Where the scores diverge sharply, you have found something more useful than a number: two senior people with incompatible beliefs about the state of the same business. That disagreement is worth more than either score.

If you want the score, it is 30 questions and 7 minutes, and nobody will ring you about it.