Organizations need their data to be available in order to make decisions, coordinate teams, and develop new artificial intelligence capabilities. However, the more information is shared, the more important it becomes to answer one question: Is each person accessing only the data they need?
A sales manager needs to know the sales figures for their territory, but not necessarily those of other regions. An operations team needs to access employee information, but not their salaries. And a user who views dashboards should not have the same permissions as someone who develops and administers the platform.
The goal is not to prevent data from circulating; it is to ensure that it circulates securely
In our article on data governance with Microsoft Fabric, we explained how to build a reliable, traceable, and business-ready data foundation. In this article, we take a closer look at one of its pillars: how to protect that data using identities, roles, and access controls that accompany the data from storage through consumption.
The usual starting point: permissions that grow without a strategy
When we review the security of a data platform, the challenges are not usually limited to technology. Often, the problem lies in how access has been granted and maintained over time:
- Excessive permissions. Users who only need to view reports also have the ability to modify content or access source data.
- Individual access rights that are difficult to manage. Authorizations are assigned on a one-to-one basis, without a clear structure of groups and responsibilities.
- Sensitive information not segmented. Personal, financial, or commercial data is shared with more user profiles than necessary.
- Protection limited to the report. Fields are hidden or visual filters are applied, but actual access to the information is not restricted.
- Lack of review. Permissions remain active even if a person changes departments or stops participating in a project.
The consequences can range from accidental modification to the exposure of confidential information. Furthermore, this situation makes it difficult to extend analytics to new teams and to undertake AI projects with confidence.
Microsoft Fabric: Multiple Layers of Security, One Goal
Microsoft Fabric brings together integration, storage, transformation, and analytics within a single ecosystem. This integration makes it easier to design a coordinated security strategy, supported by Microsoft Entra ID and complemented by the governance and protection capabilities of Microsoft Purview.
However, there is no single permission that covers all scenarios. Security is built by combining different controls:
- Identity: Who accesses the platform.
- Roles and permissions: what actions they can perform.
- Data security: which records, columns, or objects they can view.
- Governance and oversight: how information is classified and its use is monitored.
The combination of these layers allows for the application of the principle of least privilege: each person receives only the access necessary for their work, and nothing more.
Identity and Roles: The First Line of Defense
Microsoft Entra ID: A Common Corporate Identity
Fabric uses Microsoft Entra ID to authenticate users. This allows the platform to integrate with the organization’s identity policies, including multi-factor authentication and conditional access, depending on the configuration and available licenses.
For example, a company may require additional verification or a device that complies with its policies before granting access to corporate information.
The value lies in maintaining consistent identity management, rather than creating a separate user system for each tool.
Roles: Separating Administration, Development, and Consumption
Fabric workspaces have four main roles: Administrator, Member, Contributor, and Reader. These roles differentiate responsibilities for managing, creating, and viewing content.
In addition, there are specific permissions for elements such as reports, semantic models, and data warehouses.
In practice, this allows for a separation between those who build the platform and those who use its results. To distribute dashboards to a broad audience, Power BI applications also offer a consumption experience without requiring all users to be added to the development workspace.
Managing this access through Microsoft Entra ID groups makes maintenance easier: when a person changes roles, their permissions can be adjusted based on their membership in the corresponding groups.
RLS: A Single Report, Data for Each User
Row-level security, known as RLS (Row-Level Security), allows you to restrict the records that each person can view.
Let’s imagine a company operating in several regions. In a single dashboard:
- Each regional manager views sales for their region.
- Area directors access the regions under their responsibility.
- Senior management has an overall view.
There’s no need to create a separate report for each user. The content adapts to authorized access while maintaining the same metrics and business criteria.
In Fabric, RLS can be applied at different layers—such as Power BI semantic models or the data warehouse—depending on the architecture and access method.
The benefit is twofold: it protects information and reduces report duplication.
It is important to distinguish this protection from a visual filter. A filter organizes what a report displays; RLS determines which data the user is authorized to view. Furthermore, in Power BI, these restrictions are consumer-oriented: they do not apply in the same way to those with model editing permissions.
Column-level security: Viewing a record without seeing all its details
There are situations where restricting rows is not enough. A user may need to view information about an employee, a customer, or a transaction, but not all of their attributes.
For example, a planning team may need to work with employees’ departments and availability without needing access to their salaries.
Column-level security allows you to restrict access to specific fields. In Fabric Warehouse, this can be managed through permissions on table columns.
This way, a single database can meet different needs without exposing all the information to every user role.
OLS and Object Permissions: Protecting Entire Tables and Structures
Object-level security, known as OLS (Object-Level Security), allows you to restrict entire tables or columns in Power BI semantic models, including their metadata.
A model may contain both business and financial information, but certain users should only have access to the business portion. OLS allows you to establish that separation within the shared model.
At the SQL layer, object permissions also allow you to control access to structures such as tables and views, distinguishing, for example, between querying data and modifying it.
It’s important to emphasize one point here: hiding a column in a report or model is not the same as protecting it. Security must be based on effective permissions, not solely on how the content is presented.
Protecting Data from Bronze to the Dashboard
In a Medallion architecture, security must be integrated across the Bronze, Silver, and Gold layers.
Bronze: Restricted Access to Raw Data
The Bronze layer may contain information exactly as it arrives from source systems, including sensitive data that has not yet been cleaned. Access to this data must be limited primarily to ingestion processes and authorized technical profiles.
Silver: Transformation with Controlled Access
In the Silver layer, data is cleaned and standardized. Depending on requirements, unnecessary fields may also be removed, or pseudonymization techniques may be applied. Access must align with the responsibilities of the teams working on this layer.
Gold: Information Prepared for Secure Use
The Gold layer organizes data for the business. Semantic models are built on this layer, and the necessary controls are applied so that each audience can access only the information intended for them.
The mechanisms available in OneLake, Lakehouse, and Warehouse depend on the resource and the access method. Therefore, it is essential to verify protection from both Power BI and SQL, as well as from notebooks or other authorized tools.
A restriction in the report alone does not protect direct access to the source.
Microsoft Purview: Understanding Data Sensitivity and Monitoring Its Use
Permissions determine who can access data. Microsoft Purview complements this view by helping to identify and govern the information being protected.
Depending on the enabled capabilities, its integration with Fabric allows you to work with:
- Sensitivity labels, to identify confidential information.
- Catalog and lineage, to identify data assets and understand their origin.
- Auditing, to review activities and support monitoring tasks.
- Data loss prevention, for supported items and scenarios.
Labels do not replace access permissions, and the scope of protections depends on the type of content and how it is used or exported.
Similarly, having these tools helps address regulatory obligations, but does not in itself guarantee compliance: it must be accompanied by policies, responsibilities, and procedures.
How We Approach Security in a Fabric Project
At Intelequia, we incorporate security right from the solution design phase—not as a configuration step at the end of the project.
The process begins by identifying sensitive data and the user profiles that need to access it. Next, we define an access matrix, separate the responsibilities of administration, development, and end-user access, and apply the appropriate controls at each layer.
Before going live, we validate access using representative user profiles: we not only verify that each user can view what they need but also ensure they cannot access what must remain restricted.
Security continues after deployment through permission reviews, activity monitoring, and adjustments in response to organizational changes or new data sources.
Share with confidence, grow with control
The value of Microsoft Fabric lies not only in bringing data together and accelerating analysis. It also lies in providing capabilities to expand its use while maintaining control over the information.
A well-designed strategy enables more teams to work from a common foundation, reduces the need to distribute separate files, and prepares the platform for new analytics and artificial intelligence scenarios.
Greater access to knowledge doesn't necessarily mean greater data exposure.
Does your organization have an access model that's ready to scale? At Intelequia, we help you design and review the security of your Microsoft Fabric platform—from identities and roles to data protection and data consumption in Power BI.