Skip to content
Technology transformation expertise firm · ARDIANE Group Innovate, secure, succeed

Data

Where to start with a data governance strategy

In brief

A data governance strategy sets the uses to be served, the order in which data domains are taken on, the people accountable for them, their resources and how the effect obtained is measured. It begins with a single domain, chosen on explicit criteria: Management Square details these criteria here, what the legal texts require and the cases where it is better to postpone.

Page contents

On paper, data governance strategies look alike: roles, a body, a catalogue, policies. What separates them a year later comes down to where each one started.

A data governance strategy is the short document by which senior management decides which uses governance must serve, in what order data domains will be taken on, who will be accountable for them and with what resources. To the question “where to start”, the answer varies little: with a single domain, chosen because a use is waiting for it and a business owner agrees to be accountable for it. The criteria for this choice are checked before any launch. They sometimes lead to another decision, that of postponing.

The questions a data governance strategy must settle

The document answers five questions. Which uses does governance serve first: a disputed management indicator, a regulatory report, a project with a deadline? In what order will data domains be taken on, and on what condition do you move to the next? Who decides in each domain, and with how much time allocated? Who settles a disagreement between two departments? How will you see that governance is producing an effect?

It leaves aside two subjects that often clutter it. The choice of a catalogue tool comes after the first decisions. A charter of general principles meets no opposition and commits no one.

The roles of owner and steward, the glossary, the catalogue and the policies are described on our data governance page. The strategy decides where to put them in place first.

Where to start: a single data domain, chosen on criteria

A data domain groups the data of a single business object: customers, products, suppliers, contracts, employees, equipment. Starting with a single domain has a practical reason. Roles are learnt by making decisions, and an organisation does not break in ten owners at once.

Criterion Question to ask Favourable sign Warning sign
A use that is waiting Which decision, report or project with a deadline depends on this data? A named sponsor is asking for a result The use is expressed in the conditional
A business owner Who has authority over the process that produces this data, and agrees to be accountable for it? One person, with time granted by their line management Several departments pass the responsibility back and forth
A bounded scope Does the domain’s critical data fit on one page? Few critical attributes, few source applications The domain touches the whole information system
Observed defects What problems are observed: discrepancies in figures, unanswered access requests, manual reworking? Specific cases that the teams can cite General dissatisfaction, with no example
Stable sources Will the applications that produce this data remain in place? No replacement under way A migration is in progress
An external requirement Does a legal text, an audit or a contract require this data to be justified? A known deadline and recipient A requirement that nobody follows

The first two criteria are eliminatory. Without a use, the approach produces documents. Without a business owner, it produces decisions that nobody applies. The other four decide between the remaining candidates.

The customer domain deserves a word of caution. It is often proposed first, because what is at stake is visible. It is also one of the most difficult: sales, customer service, finance and marketing each create or modify part of it, and none is accountable for it alone. A domain held by a single department, such as suppliers or equipment, makes it possible to break in the roles and the arbitration body. The customer domain comes next, on a reduced scope, for example customer identification alone.

Between two equivalent candidates, choose the one whose result will be seen by a senior management body: extension to the following domains depends on it.

For personal data, the starting point is not entirely free. The GDPR, Regulation (EU) 2016/679, requires in its Article 30 that a record of processing activities be kept, subject to the exception provided for in its paragraph 5. Its Article 5(2) requires the controller to be able to demonstrate compliance with the principles set out in paragraph 1. In a large organisation, this record should therefore already exist: it provides a first inventory, and the data protection officer is a contact for the approach from the framing stage.

Regulation (EU) 2022/868, known as the Data Governance Act, does not organise a company’s internal governance, despite its name. Its first article concerns the re-use of certain categories of data held by public sector bodies, data intermediation services, the registration of entities that collect and process data made available for altruistic purposes, and a European Data Innovation Board.

Regulation (EU) 2023/2854, known as the Data Act, applicable since 12 September 2025 according to its Article 50, which sets specific dates for certain obligations, may on the other hand designate a first domain. Its first article covers in particular making data relating to connected products and related services available to their users. A manufacturer concerned has there an external requirement, in the sense of the table above.

The DAMA-DMBOK, a framework published by DAMA International, serves as a common vocabulary. Its publisher states that it is neither a prescriptive standard nor a regulatory text, and that each organisation adapts it to its context. It therefore does not designate the domain to start with.

The evidence to gather before opening a second domain

Extension is decided on evidence. Four items are enough.

  • The owner has made real decisions: a definition arbitrated, an access request settled, an anomaly assigned to someone.
  • At least one rule is enforced by a technical control. An access right is configured on the data platform, or a check runs in a processing pipeline, following the practices of our data quality offer.
  • The initial use has received what it was waiting for, and its sponsor confirms it.
  • The set-up can be reused: a mandate template, a record of decisions made, a known way of referring a matter to the arbitration body.

The second domain takes up this set-up and adjusts it. In a highly decentralised organisation, the following domains open in parallel under a small number of common rules: this is the federated governance described in our entry on data mesh.

The cases where the approach should be postponed

Postponing is a legitimate decision in five situations.

When no use is waiting, launching governance “to prepare for the future” mobilises owners on questions that nobody is asking. The first project to run into data problems will provide a better starting point.

If no business owner agrees to be accountable for the data, the obstacle is a mandate from senior management, and the useful work consists of obtaining it.

When the source application is being replaced, the definitions, the name of the owner and the migration rules are decided within the migration project itself: an approach run alongside it would concern data that is bound to change. A reorganisation in progress calls for the same caution: naming owners who will change job in the meantime means starting all over again.

That leaves the case where the real need is a localised defect. Unchecked data entry in an application is corrected in that application, without an overall strategy.

Postponing suspends no legal obligation: the record of processing activities and data security continue to apply. A postponement is decided like a launch, in writing, with the condition that will end it and the name of the person responsible for noting it.

FAQ

Frequently asked questions

Another question? Ask us

What is a data governance strategy?

A data governance strategy is a short document by which senior management decides which uses governance must serve, in what order data domains will be taken on, who will be accountable for them, with how much time allocated, and how the effect will be measured. It differs from the charter, which sets out principles, and from the choice of a catalogue tool, which comes after the first decisions.

Which data domain should you start with?

With a domain where a use is waiting for a result and where a business owner agrees to be accountable for it: these two criteria are eliminatory. The scope, the existence of observed defects, the stability of the source applications and any external requirement then decide between the candidates. The customer domain, often cited first, is rarely the simplest, because several departments share responsibility for it.

Does the European Data Governance Act require internal governance?

No. According to its first article, Regulation (EU) 2022/868 lays down the conditions for the re-use of certain categories of data held by public sector bodies, a notification and supervisory framework for data intermediation services, a framework for voluntary registration of entities that collect and process data made available for altruistic purposes, and a European Data Innovation Board. It prescribes no roles, body or catalogue to a company for its own data.

Should you follow the DAMA-DMBOK to build your strategy?

The DAMA-DMBOK, published by DAMA International, provides a vocabulary and a breakdown of data management that are useful for checking that no subject has been forgotten. Its publisher presents it as a reference framework and states that it is neither a prescriptive standard nor a regulatory text. It therefore does not designate the domain to start with: that choice depends on the uses and the owners available in your organisation.

In which cases should a data governance approach be postponed?

When no use is waiting for a result, when no business owner agrees to be accountable for the data, when the source application is being replaced or when a reorganisation is redistributing responsibilities. In these situations, the approach produces documents with no effect. Legal obligations, such as the record of processing activities provided for by Article 30 of the GDPR, continue to apply during the postponement.

A project to bring to completion?

A first conversation is enough to understand your challenge, identify the right people and tell you frankly what we can bring.