Domain ownership, no common measure
Every domain decides for itself what “good” means. Without one measure that means the same thing in all of them, you have decentralised silos rather than a mesh — and nothing that rolls up to a single view.
You moved to a data mesh, and the work is distributed exactly as it should be. What is missing is one score that means the same thing in every domain, every organization and every country.
Ownership moved out to the domains, and that was the right call. What did not move with it was a common way to tell whether each domain is actually holding up its end.
Every domain decides for itself what “good” means. Without one measure that means the same thing in all of them, you have decentralised silos rather than a mesh — and nothing that rolls up to a single view.
Each domain ships its own. Nobody holds one list of what exists, who owns it, what it promises to its consumers, or what state it is in right now.
A model inherits the quality of whatever it was fed. Which data products are fit to feed one — and which are not — is now a question somebody asks you directly.
A maturity score per data product tells you what is fit to feed a model and what is not — before it reaches one, rather than after an answer embarrasses somebody.
A score that compounds makes progress visible, so data investment becomes defensible at budget time instead of being argued from anecdote.
Frameworks are optional here — load your own internal standards, or start from the built-in ones.
The same score, computed from the same evidence, rolled up wherever you need to compare. A number only means something if it means the same thing in Madrid as it does in Mexico City.
Which domains are carrying the estate and which are behind, on one scale — so the comparison is a fact rather than a negotiation.
How many data products have a named owner, a current contract and quality rules. And, just as usefully, how many do not.
Spend per data product set against what it actually delivers, by domain and by country — the root cause most organizations name first.
Time from request to live, which gates were passed on time, and where the same domain keeps slipping.
Freshness, quality results and SLA adherence per product, live — not a monthly export that is stale before it lands.
Who consumes which product, what is being requested through the marketplace, and what people keep asking for that does not exist yet.
The pilot produces these figures from your own estate. We will not put invented savings on a slide.
The picture of what exists, who owns it and what state it is in is standing rather than assembled from each domain, so it moves from weeks of senior time to a report you run.
Issues open a governed workflow when they are detected, not when the business notices — the expensive gap closes.
Each governance cycle builds on the last, so the platform you bought last year keeps compounding instead of being replaced.
Teams act on data with a known maturity score instead of debating whether it can be trusted.
analyst-days per estate report × reports per year × loaded day rate
hours from failure to discovery × incidents per year × cost per hour
governance tooling written off in the last three years ÷ 3
Nothing on this page is a different product. It is the same measurement, pointed at the questions a domain-owned estate actually raises.
“Are our data products mature enough?” becomes a score, per product, on a date — and a trend by the next quarter, because re-running it costs nothing.
Cost against value delivered, per product. The root cause most organisations name first is that nobody can show what the data investment bought.
Whoever requested a data product watches it the whole way, and sees how it is performing once it is live — without a status meeting.
Every governed object published with its owner, its score and its evidence. Access is requested there, reviewed by the owner and logged.
Bronze through Gold, earned automatically from evidence you already hold — so whoever picks a product out of the marketplace can see what it is worth before they build on it.
Each domain runs itself; the standard, the glossary and the measure stay common. That is the whole difference between a mesh and a set of silos with better branding.
A real slice of your estate, about half a day a week from one of your people, and an evidence pack you keep either way.