Introduction

In modern data-driven organisations, one of the biggest challenges is balancing accessibility of data with strong governance controls. Business users want to easily find and use trusted data, while data teams must ensure consistency, quality, and security. Microsoft Fabric addresses this challenge with a powerful combination of semantic model (SM) sharing, OneLake Catalog discoverability, and governance features, enabling organisations to scale self-service analytics with good governance practices.

The Data Accessibility vs. Governance Dilemma

The Challenge:

• Users struggle to find data they need

• Data teams lose control when access is too open

• Shadow datasets multiply due to lack of visibility

• Governance becomes difficult to enforce at scale

• The Solution: Fabric's integrated approach that separates access from control while enabling organization-wide discovery.

Decoupling Access from Control

A common misconception is that users need access to an entire workspace to consume data. In Fabric, this is not the case. Semantic models can be shared independently of the workspace, allowing:

• Users to connect via Power BI or Excel
• Access to trusted data without workspace visibility
• Centralised control over who can modify models

This aligns strongly with the principle of least privilege, where users only receive the access necessary to perform their role. When granting access to a semantic model, permissions can be carefully configured. For example:

• Allow users to build content using the model
• Restrict their ability to edit or reshare
• Prevent unintended changes to governed data assets

This ensures that business users can confidently build insights while the integrity of the underlying data remains protected.

---

Practical Example: Sales Analytics Model Scenario: Finance team needs access to a Sales Semantic Model

• Permissions granted: Read + Build (can create reports)
• Permissions restricted: Edit, Reshare, Modify model structure
• Result: Users get insights without risking data integrity

---

The image above shows the access‑granting options for a semantic model. It shows the selected user and provides three permission settings: (a) the ability to modify the model, (b) to share it, and to (c) build content using its data.

Building Trust Through Endorsement

Accessibility alone is not enough, as users also need to know which data to trust. Fabric provides endorsement mechanisms such as:

• Promoted: Indicates useful and widely adopted assets
• Certified: Highlights trusted, governed, and validated data

These endorsements surface directly in user tools (Power BI, Excel), guiding users toward high-quality data and reducing reliance on unofficial or duplicated datasets. The result is a shift from fragmented reporting to a single source of truth, where trusted semantic models become the foundation for analytics across the organisation.

How to Certify a Model

1. Navigate to the semantic model in Fabric
2. Click the three-dot menu → Endorsement
3. Select Certified and add a certification message
4. Endorsed models appear with a badge in search results and catalogs

Solving the Discoverability Challenge

One of the most common pain points in large organisations is simple:

"How do users know what data exists if they don't already have access to it?"

Fabric addresses this through discoverability, powered by the OneLake Catalog.

What is Discoverability in Microsoft Fabric?

Discoverability allows data teams to make certain assets visible across the organisation without granting access to them. When enabled:

• Users can search and browse available semantic models
• Items appear even if users do not have permission to access them
• A clear message indicates restricted access, with an option to request it This shifts the experience from "hidden data silos" to a transparent, searchable data ecosystem, particularly important in large organisations.

OneLake Catalog in Action

The OneLake Catalog acts as a centralised metadata layer across Fabric:

• A single location to explore available data assets
• Visibility into semantic models, reports, and other artefacts
• Seamless integration with governance controls

---

Example:

1. User searches for "Customer Sales" in OneLake Catalog
2. Multiple models appear, some marked "Certified"
3. User can see:
   • Model owner & contact
   • Description and business purpose
   • Endorsement status
   • Row-level security applied (yes/no)
4. If user lacks access, they can request permission
5. Data owner receives notification and approves/denies

---

Controlled Visibility with Governance

Discoverability is not an "open everything" approach, it is tightly governed. Fabric provides controls at multiple levels:

Item-Level Control

Data owners and stewards decide:

• Which semantic models are discoverable
• Which remain restricted or hidden
• Sensitive domains (e.g., workforce or confidential datasets) can remain undiscoverable, while broader datasets can be found in the catalog

How to Enable Discoverability on a Semantic Model:

• Go to the semantic model settings
• Under Endorsement & Discovery → select Make discoverable
• Add a business description explaining the model's purpose
• Specify the intended audience/use case
• Save and publish

Tenant-Level Control

Administrators can define:

• Who is allowed to mark items as discoverable
• Who can view discoverable items
• Domains that should remain hidden by default

Admin Configuration (in Fabric Admin Portal):

• Tenant settings → Information Protection
• Define who can endorse/certify models
• Set discoverability policies by department or domain
• This ensures consistency in how discoverability is applied across the organisation.

This ensures consistency in how discoverability is applied across the organisation.

Understanding Impact with Lineage

As adoption grows, understanding dependencies becomes critical. Fabric's lineage view provides:

• Visibility of how semantic models are used
• Connections between datasets, reports, and downstream assets
• Insight into which teams or workspaces depend on a model

This allows data teams to:

• Assess impact before changes or updates
• Identify widely used assets requiring extra governance
• Strengthen governance through transparency
• Plan deprecation strategies for unused models

Scenario: You need to update a critical semantic model

• Lineage shows 47 downstream reports depend on it
• You identify 3 teams affected
• You can plan updates carefully and notify stakeholders
• Result: No broken dashboards or frustrated users

Summary

By combining these capabilities, Fabric enables a modern data operating model: Centralised Governance

• Controlled semantic models with restricted editing
• Endorsement and certification workflows
• Managed visibility through discoverability settings

Decentralised Consumption

• Users build reports independently
• Access via Power BI or Excel (no workspace access needed)
• Self-service analytics at scale

Transparent Discovery

• OneLake Catalog surfaces available assets organization-wide
• Users can request access to the right data
• Reduced duplication and shadow datasets

Source:

• https://learn.microsoft.com/en-gb/power-bi/collaborate-share/service-discovery
• https://learn.microsoft.com/en-gb/power-bi/collaborate-share/service-discovery#how-to-set-up-semantic-model-discovery-in-an-organization
• https://learn.microsoft.com/en-gb/fabric/governance/onelake-catalog-overview?toc=%2Fpower-bi%2Fconnect-data%2Ftoc.json&bc=%2Fpower-bi%2Fconnect-data%2Fbreadcrumb%2Ftoc.json
• https://learn.microsoft.com/en-gb/power-bi/collaborate-share/service-discovery#how-to-set-up-semantic-model-discovery-in-an-organization

Disclaimer: This content was accurate at the time of initial publication. Microsoft Fabric features and capabilities may have evolved since then.