How business and technical metadata, definitions, classifications, and lineage are documented, standardised, and kept discoverable and accurate.
MDT-CON-01
Understanding of business terms
When you encounter a business term or measure in a report or dataset, how easily can you find an agreed, authoritative definition of what it actually means?
Maturity level descriptions
No business glossary or definitions are available to the Data Consumer; the meaning of terms is inferred from context or assumed. The Data Consumer interprets terms such as “active customer” or “completed case” by assumption, with no definition available anywhere.
Definitions exist in scattered documents or in individuals’ heads; the Data Consumer must ask a colleague and answers may differ. Different colleagues give different definitions of the same term, and nothing authoritative resolves the difference.
An agreed business glossary exists in a defined, accessible location covering priority terms, and business users are shown how to use it. The Data Consumer can look up an authoritative definition for the terms they most commonly encounter.
Glossary coverage and usage are measured, definitions are formally owned and reviewed on a defined cycle, and terms are linked to the data assets that implement them. The Data Consumer finds definitions for essentially all terms they use, each with a named owner and review date, linked to the reports and fields concerned.
Definitions are surfaced in context wherever the term appears and are maintained continuously, so the Data Consumer never needs to leave their work to establish meaning. The Data Consumer sees the authoritative definition directly in the report or tool when they hover over or select a measure.
MDT-CON-02
Discoverability of data assets
When you need data for a new question, how easily can you find out what data and reports already exist across the organisation, without relying on personal networks?
Maturity level descriptions
No means exists to discover what data or reports exist; the Data Consumer finds out only through colleagues or by chance. Finding relevant data depends entirely on knowing the right person to ask, and comparable assets are unknowingly rebuilt.
Partial listings exist (a reports index, a shared folder), incomplete and out of date, so the Data Consumer cannot rely on them. The Data Consumer consults a list that is missing much of what exists and includes items that no longer work.
A searchable catalogue of data assets exists covering priority data, is accessible to business users, and they are trained to use it. The Data Consumer can search a catalogue by business term and find relevant datasets and reports with a description of each.
Catalogue coverage, search success and adoption are measured, with gaps actively closed and business-user search behaviour used to improve the catalogue. The Data Consumer usually finds what they need first time, and the organisation measures search success and closes discovered gaps.
Discovery is proactive and personalised — relevant assets are recommended to the Data Consumer based on role and activity, and coverage is maintained automatically. The Data Consumer is offered relevant certified assets proactively, and new assets appear in the catalogue automatically as they are created.
MDT-CON-03
Interpretive context and caveats
Beyond definitions, how well are you given the context you need to interpret a dataset correctly — its coverage, collection method, known limitations and the situations it should not be used for?
Maturity level descriptions
No interpretive context is provided; the Data Consumer sees only field names and values, with no guidance on correct interpretation. Data is used without knowing how it was collected, what it excludes, or where it is unreliable.
Context is available only by asking someone knowledgeable, and is not recorded; caveats are transmitted by word of mouth if at all. The Data Consumer learns a dataset’s limitations only after misusing it, or from a colleague who happens to know.
Documented context — coverage, collection method, known limitations and unsuitable uses — accompanies priority datasets in an accessible location. The Data Consumer can read a written statement of what a dataset covers, how it was collected and what it should not be used for.
Context documentation is maintained on a defined cycle, its completeness is measured, and caveats are actively surfaced to users of affected assets. The Data Consumer is alerted to a newly identified limitation on a dataset they use, rather than discovering it independently.
Interpretive guidance is embedded at the point of use and adapts to how the asset is being used, with misuse actively discouraged in context. The Data Consumer is warned within the tool when they use a dataset in a way its documented limitations do not support.
MDT-CON-04
Consistency of definitions
When the same business concept appears in different systems, reports or business areas, how consistently is it defined and named, as opposed to carrying conflicting meanings?
Maturity level descriptions
Definitions conflict freely across the organisation; the Data Consumer regularly finds the same term meaning different things in different places. The same measure name produces incompatible figures in different reports, with no way to reconcile them.
Inconsistencies are recognised anecdotally and reconciled manually each time they cause a problem, with no lasting resolution. The Data Consumer or an analyst reconciles conflicting definitions by hand whenever a discrepancy is challenged.
Organisation-wide standard definitions are agreed for priority concepts and published, with conflicting local variants identified. The Data Consumer can identify which definition is the organisational standard and where a local variant differs from it.
Conformance to standard definitions is tracked across business-facing assets, with non-conforming variants scheduled for remediation. The Data Consumer increasingly finds the same figure defined identically across reports, with remaining exceptions documented and being addressed.
Standard definitions are enforced through a shared semantic layer, so consistent meaning is applied automatically wherever the concept is used. The Data Consumer’s reports draw measures from a single governed semantic definition, making divergent definitions structurally difficult to create.
MDT-CON-05
Trust in metadata currency
How confident can you be that the documentation describing the data you use is current and accurate, rather than describing how things used to be?
Maturity level descriptions
Documentation, where it exists, is known to be unreliable; the Data Consumer disregards it and asks a person instead. Documentation describes fields and reports that no longer exist, so it is not consulted.
Some documentation is current and some is not, with no way for the Data Consumer to tell which is which. The Data Consumer cannot distinguish a recently maintained entry from an abandoned one, so trusts none of it fully.
Documentation carries ownership and last-reviewed information, and priority assets are reviewed on a defined cycle. The Data Consumer can see who owns an entry and when it was last reviewed, and priority entries are demonstrably current.
Currency is measured across the documented estate, stale entries are flagged, and completion of review is tracked and reported. The Data Consumer sees stale entries clearly marked, and the organisation reports on how much documentation is current.
Technical metadata is harvested automatically and business metadata review is triggered by change, so documentation cannot silently fall out of date. The Data Consumer relies on documentation as a matter of course because a change to the data automatically updates or flags the corresponding entry.
MDT-CON-06
Sensitivity and permitted use metadata
How clearly can you tell, from the data asset itself, how sensitive it is and what you are permitted to do with it?
Maturity level descriptions
No sensitivity or permitted-use information accompanies data assets; the Data Consumer judges sensitivity by intuition. The Data Consumer decides for themselves whether a dataset containing personal information may be shared or exported.
Sensitivity is sometimes communicated when data is provided, inconsistently and not recorded against the asset. The Data Consumer is occasionally told a dataset is sensitive at the point of handover, with nothing recorded on the asset itself.
Data assets carry a documented classification and statement of permitted use, visible to business users at the point of access. The Data Consumer can see a classification label and permitted-use statement on the datasets and reports they access.
Classification coverage is measured and maintained, reviewed on a defined cycle, and drives the handling controls applied to business users. The Data Consumer finds current classifications on essentially all assets they use, and those classifications visibly determine what they can do with them.
Classification is applied and maintained automatically, propagates to derived assets, and enforces permitted use at the point of action. The Data Consumer’s derived report inherits the sensitivity of its source automatically, and controls act on that classification without manual intervention.
MDT-CON-07
Consumer contribution to metadata
How effectively can you contribute to the organisation’s metadata — flagging an unclear definition, requesting a missing one, rating an asset, or adding usage context others would benefit from?
Maturity level descriptions
No mechanism exists for business users to contribute to or correct metadata; the Data Consumer’s knowledge stays with them. The Data Consumer knows a published definition is wrong but has no way of saying so.
Corrections can be raised informally by contacting a person, with no process and no visibility of whether anything changed. The Data Consumer emails someone about an incorrect definition and never learns whether it was updated.
A defined route exists for business users to request, comment on or correct metadata, and it is communicated to them. The Data Consumer can submit a definition request or correction through a named mechanism and it reaches an owner.
Contributions are tracked to outcome, with response expectations, and business-user contribution levels are measured and encouraged. The Data Consumer receives an outcome on submitted corrections, and contribution volumes are monitored as an engagement measure.
Metadata is genuinely collaborative — business users annotate, rate and endorse assets directly, with curation ensuring quality and contributions visibly improving the catalogue. The Data Consumer adds usage notes and ratings directly to catalogue entries, and other users benefit from that context.
MDT-CON-08
Metadata for AI-related assets
For AI-assisted outputs and AI-supported datasets you use, how well are you told what they were built from, what their limitations are, and what they may be relied on for?
Maturity level descriptions
No information accompanies AI-related outputs or assets; the Data Consumer cannot tell what an AI answer was based on or how far to trust it. AI-generated content arrives with no provenance, no limitations and no statement of intended use.
General caution is communicated informally, but nothing asset-specific describes sources, limitations or intended use. The Data Consumer is told broadly to be careful with AI output, with no information specific to the tool or dataset in front of them.
Documented information accompanies AI-supported assets covering data sources, known limitations and intended use, accessible to business users. The Data Consumer can consult a statement describing what an AI assistant draws on, what it is intended for and where it is unreliable.
AI asset documentation is maintained on a defined cycle, its coverage measured, and material changes to sources or limitations are communicated to affected users. The Data Consumer is informed when an AI tool’s underlying sources or known limitations change materially.
Provenance and limitation information is surfaced automatically with each AI output, showing the specific sources used and applicable caveats in context. The Data Consumer sees, alongside each AI answer, the specific documents or datasets it drew on and the caveats that apply to them.