Data & AI Readiness
Twelve questions across the four things that decide whether AI works on your data: can you reach it, do you agree what it means, does anyone own it, and can the business actually use it. Answer honestly — the score is only useful if it is uncomfortable. Nothing is saved, nothing is sent, and there is no sign-up.
Twelve questions
Pick the answer closest to the truth today, not the plan.
Q1How much of the data behind your key reports is connected rather than exported by hand?
Q2How long does it take to add a new source system?
Q3Can you see what changed upstream before it breaks a report?
Q4If two teams report revenue, do they get the same number?
Q5Where do metric definitions live?
Q6Is there one agreed view of each customer, supplier and product?
Q7Does each important data element have a named owner?
Q8Can you trace a figure in the board pack back to its source entry?
Q9How do you know the data is fit to use?
Q10How long from a business question to an answer people will act on?
Q11Are AI tools answering from your governed data, or from whatever they can reach?
Q12Does the data change what people do, or only what they read?
Your readiness
Updates as you answer.
Scoring is even across the four dimensions, which is deliberate: the weakest one usually sets the ceiling. A high connect score with a low define score is the most common pattern, and the most frustrating — the data is all there and nobody trusts the answer.
Want the version with your own data?
We will walk your estate and show what each dimension looks like in practice.
01 The four dimensions
Why these four, in this order
They build on each other. You cannot agree what a number means before you can reach the data, and governing a definition nobody uses achieves nothing. Skipping a step is the most reliable way to spend two years and end up with a warehouse people do not trust.
Connect
Can you reach the data where it already lives — ERP, CRM, MES, treasury, external feeds — without a project for each one, and do you find out when something upstream changes?
Next: Enterprise Data Hub · Data Integration trackDefine
Does each number mean one thing? This is where most estates stall: the data is available and two teams still produce two answers, because the definition was never written down anywhere a tool could read it.
Next: Semantic Modelling · Metric DictionaryGovern
Owners, quality rules and lineage. Not a committee — a named person per critical element, automated checks, and the ability to trace any figure back to the entry it came from.
Next: Data Governance · Governance trackActivate
Whether any of it changes what people do. Fast answers, decisions made from the same numbers, and AI that reads governed metrics instead of guessing at raw tables.
Next: Agentic AI track · AI Copilot02 What to do with the score
Fix the lowest one first
The temptation is to push the dimension you are already good at, because progress feels faster there. It rarely helps. If governance is at 30% and connectivity at 85%, more pipelines just deliver more numbers nobody trusts.
Pick the weakest dimension, pick one domain inside it — finance, or supply chain, not the whole estate — and get that one to the next band. A narrow, finished thing is worth more than a broad, half-finished one, and it gives you the evidence to fund the next step.