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

Data

A data governance policy is judged by the arbitrations it writes down

In brief

A data governance policy is the text that sets the lasting rules on data: scope, roles, classification, access, arbitration of definitions, handling of gaps. Management Square suggests leaving operating procedures to separate documents and having it approved by executive management or the executive committee; even once approved, it does not on its own establish compliance with a regulation.

Page contents

On the day two departments calculate the same indicator differently, or when a project asks for a table that its owner refuses to open, the data governance policy is consulted: it designates a rule and a person to decide, or it says nothing. General principles do not resolve such a case. Nor does a tool’s user guide, outdated at the first version upgrade.

A data governance policy is the text that sets the lasting rules: the data to which it applies, the people who decide, the way a data item is classified and opened up, the way a definition is arbitrated and a gap is handled. Operating procedures go into separate documents, revised at their own pace. This split is a methodological proposal, as is the level of approval we adopt: executive management or the executive committee. As for compliance with a regulation, an approved policy does not establish it by itself.

The data governance policy among neighbouring texts

Vocabulary is a matter of convention, and the convention in this article separates four objects. Strategy decides which domain to start with and in what order to extend: this is the subject of our article on where to start a data governance strategy. The charter sets out principles. The policy sets rules and says who enforces them. The procedure describes actions.

Our page on data governance also distinguishes four families of data policies: classification, access, retention, sharing with third parties. The governance policy is their common text. It sets out the roles and decision-making mechanisms on which these thematic rules depend, and can contain all of them when the data estate is not very extensive.

The points on which the text takes a position

A useful rule can be recognised by this: placed before a case, it gives an answer or designates the person who will give it. The table contrasts, for each point, a vague formula with what the text gains by writing down.

Point to settle Formula that leaves everything open Wording that decides
Scope “All of the company’s data” The domains covered at the date of approval, the entities concerned, what happens to data received from third parties
Roles “Each data item has an owner” Who appoints the owners, by what act, with what time set aside, and who decides in their absence
Classification and access “Sensitive data are protected” The classes adopted, the class of a data item not yet classified, the person who authorises access for each class
Arbitration of definitions “Definitions are harmonised” The forum to which the matter is referred when two domains disagree, the time within which it responds, the scope of its decision
Handling of gaps “Compliance with the policy is checked” Who records a gap, who grants an exemption, for how long, and where it is recorded

The scope is written with its exclusions. A text that states it does not yet cover unstructured documents, or a given subsidiary, is more accurate than one that claims to govern everything.

For roles, the policy sets the method of appointment and leaves the names to a register. The split between owner, steward and technical custodian is detailed in our comparison of data owner and data steward.

The default class carries more weight than it seems. On the day of approval, some of the data has not received any class: the rule says whether it is open or closed in the meantime, and who removes the doubt.

The rules on exemptions avoid two pitfalls: the rule that everyone silently gets around, and the one that blocks a legitimate project. An exemption granted bears a reason, an expiry date and a signatory.

Operating procedures referred to separate documents

The text is approved high up in the organisation: it must be able to remain stable when tools change. One criterion helps to sort. A sentence that would become false with a change of tool, organisation chart or volume belongs in a procedure. The following therefore leave the text:

  • the path of an access request: form, steps, tool, time for each step;
  • maintaining the glossary and the data catalogue;
  • the circuit for handling an anomaly;
  • retention periods by category of data, kept in a reference repository;
  • the named register of owners and stewards.

The policy cites each of these procedures and names the function that maintains it. It also sets what a procedure cannot change, for example the capacity of the person who authorises access.

For personal data, the record of processing activities provided for in Article 30 of the GDPR, Regulation (EU) 2016/679, read on 12 November 2026 in its consolidated version, and impact assessments keep their own documentation. The policy refers to them without reproducing them.

Approval at the level that commits all departments

The text sets aside time for business managers and gives a forum the power to decide between departments. Approved by the data function alone, it would commit only that function. This is why we suggest approval by executive management or by the executive committee, on presentation by the chief data officer or the function that runs governance.

The draft is prepared with the people who will have to apply it: owners already in post, the IT department, security, the legal function and the data protection officer. Article 38(1) of the same regulation requires the latter to be involved, properly and in a timely manner, in all issues which relate to the protection of personal data. Article 24(2) provides that the controller’s measures shall include, where proportionate, the implementation of appropriate data protection policies. Whether your governance policy counts among these policies is a question for the organisation’s data protection officer and lawyers.

Sector-specific texts carry their own expectations. For the banks they cover, the Basel Committee principles presented in our entry on BCBS 239 are one example. The policy can help to meet them, without being enough to demonstrate that they are complied with.

Revising the text, or waiting to write it

When the text exists, its revision is triggered by events: the opening of a domain, a reorganisation, a change in the applicable regulation. Between two revisions, its application can be read in the number of exemptions granted, in the referrals to the arbitration forum and the time taken to respond to them, and in the share of rules that a technical control enforces on the platform. A rule that accumulates exemptions may be badly written or badly placed: this count singles it out for the next revision.

When it has yet to be written, the question is one of timing. A policy drafted before any practice codifies assumptions. We suggest writing it after a first domain has made real decisions, on the basis of the log of those decisions: the rules recorded are then those that have already resolved a case. If an external deadline requires a text earlier, a version reduced to the roles, the arbitration forum and the default class can be approved, with the date on which it will be taken up again.

FAQ

Frequently asked questions

Another question? Ask us

What is the difference between a data governance policy and a data governance charter?

By convention, a charter sets out principles and a policy sets rules: who decides, how a data item is classified, who authorises access, how a disagreement is settled. Titles vary from one organisation to another, and a text called a charter may contain real rules. The criterion is the content: a text that does not make it possible to resolve any concrete case remains a statement of principles, whatever its title.

Is a data governance policy mandatory?

For personal data, Article 24(2) of the GDPR, Regulation (EU) 2016/679, provides that the controller's measures shall include, where proportionate in relation to processing activities, the implementation of appropriate data protection policies. This text does not name the data governance policy, which covers a broader field. The extent of a given organisation's obligations, under this regulation or a sector-specific text, is assessed with its lawyers.

Should there be a single policy or one policy per data domain?

We suggest a single common text, which sets the roles, the classes, the arbitration forum and the rules on exemptions, supplemented by domain annexes. Each annex lists the domain's critical data and the rules specific to it, without being able to contradict the common text. In a group, the same logic separates the group policy from each entity's adaptations, approved locally.

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.