Governance and administration

Power BI Row Level Security Setup: A Head of Data Guide

A plain guide to Power BI row level security setup: roles, DAX rules, dynamic filters, testing, and the mistakes that leak data across teams.

Austin Levine

·

Aktualisiert

A Power BI row level security setup means you create roles in Power BI Desktop, write a DAX filter on the tables that store sensitive rows, then assign people to those roles in the Power BI Service. Done well, a sales manager sees only their region and a finance user sees only their entity, all from one shared report. The work that trips teams up isn't the roles, it's testing them and keeping them in sync as people and reports change.

A regional manager opens the company sales report and sees every other region's numbers, including deals that were never meant to leave head office. That problem is exactly what a Power BI row level security setup is supposed to prevent, and it's the reason a Head of Data gets a nervous message on a Monday morning. The report was shared with the whole company, the data model has every region in one table, and nothing stops one person from reading another team's rows. You don't want a separate report per region. You want one report where each person sees only what they're allowed to see.

This guide walks through how row level security actually works in Power BI, where teams get it wrong, and how to test it before it ships.

Key takeaways

  • Row-level security creates roles in Desktop, writes a DAX filter, then assigns people in the service.

  • Done well, a sales manager sees only their region from the same shared report.

  • Most teams start with static roles and outgrow them, moving to dynamic roles that scale.

  • Testing is the step people skip, and it is the step that saves the Monday-morning message.

What's row level security in Power BI?

Row level security, or RLS, filters the rows a person can read inside a report, based on who they are. The report, the visuals, and the data model stay the same for everyone. What changes is the slice of rows each person is allowed to see.

It works through two pieces. First, a role, which is a named rule you build in Power BI Desktop. Second, a DAX filter on a table, which is the logic that decides which rows that role keeps. A role called "Germany" might carry the rule `[Region] = "Germany"`. Anyone in that role sees German rows and nothing else. The person still opens the same report link as everyone else. They just get a filtered view.

One thing to be clear about: RLS filters data, it doesn't hide report pages, visuals, or measures. If you need to hide a whole page from a group, that's a different job. RLS is about rows.

How do you do a Power BI row level security setup, step by step?

Here's the core setup, from an empty model to a working filter.

  1. Open your report in Power BI Desktop and go to the Modeling tab. Choose Manage roles.

  2. Create a role and give it a plain name, like "Sales Rep" or "Region Manager".

  3. Pick the table that stores the sensitive rows, and write a DAX rule on it. For a fixed role that's a simple filter, for example `[Region] = "West"`.

  4. Save the role, then use View as roles to preview the report as that role before you publish anything.

  5. Publish the report to the Power BI Service.

  6. In the Service, open the dataset settings, find Security, and add people or security groups to each role.

That last step matters. Roles created in Desktop are empty until you assign members in the Service. A role with no members filters nobody, so the rule sits there doing nothing and the data stays open. Assigning people, ideally through Microsoft Entra security groups rather than one name at a time, is where the setup becomes real.

Static or dynamic: which RLS roles should you use?

Most teams start with static roles and outgrow them fast. The two approaches solve the same problem at very different scales.

Static RLS roles

Dynamic RLS roles

How it filters

A fixed value written into the rule, like Region = West

DAX used

Simple column filter

Good for

A handful of regions or entities that rarely change

The cost

A new role every time the org changes

Static roles are fine when you have five regions and a stable org. The moment you have two hundred sales reps and access that shifts every month, a role per person is unworkable. Dynamic row level security fixes that. You build one role that reads the email of the person viewing the report, using `USERPRINCIPALNAME()`, and matches it against a table that lists which rows each email is allowed to see. Add a new rep to the mapping table and they're covered, no new role, no republish.

The catch with dynamic RLS is that your permissions table becomes a live part of your governance. If it's wrong, access is wrong, and nobody notices until someone sees too much.

Why do row level security setups leak data?

Gartner expects 80% of data-governance initiatives to miss their goals by 2027, and RLS is one of the quiet places that gap shows up. The rules look fine in a demo and leak in production. A few reasons keep coming back.

The role has no members, so it never filters anyone. The relationship between the fact table and the filtered table points the wrong way, so the filter never travels to the rows that matter. A workspace admin or dataset owner is exempt from RLS entirely, which is correct behavior but a nasty surprise if you test as an admin and declare it safe. And the classic one: someone edits the model months later, adds a table, and the new table has no rule, so a whole set of sensitive columns walks straight past the roles you built.

None of these are exotic. They're the normal result of a setup that was never tested as a real end user.

How do you test row level security before it ships?

Testing is the step people skip, and it's the step that saves the Monday-morning message. Do it in two passes.

In Power BI Desktop, use View as roles to preview the report as each role, and confirm the row counts drop to what that role should see. Then, in the Power BI Service, open the dataset security page and use Test as role to view the live report exactly as a member of that role would. Check a real total, not just that the page loads. If the "West" role still shows a national number somewhere, a measure or a table is escaping the filter.

Don't test as yourself if you're the dataset owner, because owners bypass RLS. Ask one real person from each group to open the report and tell you the single number they see, then confirm it matches what they're allowed to see. That five-minute check is worth more than any amount of staring at DAX.

If you want a faster read on where your reports stand, our free Power BI Governance and Cost Scorecard gives you a five-minute assessment of your access, ownership, and refresh risks, with the top three things to fix first. If the rules still aren't working, book a governance assessment with us and we'll pressure-test your row level security setup on your real reports.

FAQ

Does row level security slow down a Power BI report? It can, mostly with dynamic RLS on large models, because the filter runs on every query. Keep the permissions table small and well-related to the fact table, avoid heavy DAX inside the role rule, and test refresh and load times with the role applied, not just as an open report.

Can an admin or report owner see everything even with RLS on? Yes. Dataset owners, workspace admins, and members with edit rights bypass row level security by design. That's why you must test as a normal role member in the Service, and why an admin confirming "looks filtered to me" isn't a valid test.

Does row level security work with Power BI Copilot and Q&A? Yes, RLS filters carry into Q&A and Copilot answers, so a user only gets results from rows their role allows. That makes your row level security setup part of AI governance too, because a wrong permissions table can feed the wrong numbers into an AI answer.

Read next: Power BI governance framework: the pillars that keep numbers trusted and Power BI workspace roles and permissions explained.

Ist Ihr Power BI richtig aufgesetzt?

1

Buchen Sie 45 Minuten und sagen Sie uns, worum es geht.

2

Wir gehen Ihre Berichte und Ihre Umgebung live mit Ihnen durch.

3

Innerhalb von zwei Arbeitstagen erhalten Sie unseren Plan schriftlich.

Ist Ihr Power BI richtig aufgesetzt?

1

Buchen Sie 45 Minuten und sagen Sie uns, worum es geht.

2

Wir gehen Ihre Berichte und Ihre Umgebung live mit Ihnen durch.

3

Innerhalb von zwei Arbeitstagen erhalten Sie unseren Plan schriftlich.

Ist Ihr Power BI richtig aufgesetzt?

1

Buchen Sie 45 Minuten und sagen Sie uns, worum es geht.

2

Wir gehen Ihre Berichte und Ihre Umgebung live mit Ihnen durch.

3

Innerhalb von zwei Arbeitstagen erhalten Sie unseren Plan schriftlich.

CaseWhen ist eine BI-Beratung aus Berlin. Wir bauen Reporting, dem die Geschäftsführung vertraut, auf dem Microsoft-Stack: Power BI, Fabric und Azure.

Berlin, Deutschland

© CaseWhen Consulting GmbH

Deutsch

CaseWhen ist eine BI-Beratung aus Berlin. Wir bauen Reporting, dem die Geschäftsführung vertraut, auf dem Microsoft-Stack: Power BI, Fabric und Azure.

Berlin, Deutschland

© CaseWhen Consulting GmbH

Deutsch

CaseWhen ist eine BI-Beratung aus Berlin. Wir bauen Reporting, dem die Geschäftsführung vertraut, auf dem Microsoft-Stack: Power BI, Fabric und Azure.

Berlin, Deutschland

© CaseWhen Consulting GmbH

Deutsch