Power BI decomposition tree: break one number down, level by level
Power BI decomposition tree is a visual that breaks a single measure down across several dimensions. A reader starts with one total, then splits it by any field in any order. The visual can also suggest the next split that gives the highest or lowest value. This guide covers the setup, the suggested splits, locked levels and the limits that Microsoft documents. In short, put one measure in Analyze and several fields in Explain by, then select the plus sign on a node to choose the next split.
- Drill in any order
- High and low value splits
- Locked levels
Last updated
What is the Power BI decomposition tree?
Microsoft describes it as a way to visualize data across multiple dimensions. The visual aggregates the data automatically and lets you drill into the dimensions in any order. Therefore it suits ad hoc exploration and root-cause analysis.
A normal drill path follows a fixed hierarchy. This visual has no fixed order. One reader may split backorders by product type first, while another starts with the plant. Both reach an answer without asking the author for a new chart.
The tree also works with the rest of the page. When a reader selects a value in another visual, that selection cross-filters the tree, and the order of its nodes can change. Selecting a node on the last level cross-filters the other visuals in turn.
Build a Power BI decomposition tree in four steps
The steps follow Microsoft's tutorial and work in Power BI Desktop as well as in the Power BI service.
Add the visual
First select the decomposition tree icon in the Visualizations pane. An empty visual then appears.
Set the metric
Next drag a measure or an aggregate into Analyze. The visual then shows the total as the root node.
Add the dimensions
Then drag one or more fields into Explain by. A plus sign appears next to the root node as a result.
Split the node
Finally select the plus sign and choose a field. The tree then expands, and you can repeat this on any node.
How do the high value and low value splits work?
Two options sit at the top of the list when you select a plus sign. High value considers all available fields and picks the one that gives the highest value of the measure. Low value does the same for the lowest value.
A light bulb marks a level that the visual chose this way. Hover over it, and a tooltip explains the finding in a sentence. These levels also recalculate when you select a different node or cross-filter the tree, so a different field may appear.
You can change the analysis from absolute to relative. Relative mode looks for a value that stands out against the others in its column, instead of the largest figure. However, one rule applies to the order: a manual level cannot follow a suggested one without turning it into a normal level.
Three author controls for a guided analysis
A report author can shape how much freedom readers get, and Microsoft documents these controls.
Locked levels
An author can lock a level so readers cannot remove or change it. They can still explore different paths inside it.
Lock order
You can lock as many levels as you want. However, an unlocked level cannot come before a locked one.
Suggested splits off
You can turn off the high and low value options in the analysis formatting settings, so readers split by named fields only.
Which limits apply to the Power BI decomposition tree?
Microsoft documents three numeric limits. The tree allows at most 50 levels. It can show at most 5,000 data points at one time. Also, each level shows only its top 10 values at present.
Source support differs too. The visual does not work with on-premises Analysis Services. The suggested splits do not work with Azure Analysis Services, Power BI Report Server or Publish to web. They also do not work with complex measures in Analyze.
So check the connection type before you promise this visual to a team. When a model uses one of those sources, a matrix with a clear hierarchy may be the safer choice.
Good habits for a tree that readers can follow
A tree grows wide quickly, so a little preparation keeps the analysis readable.
- Choose one measure with a clear business meaning
- Offer a short list of dimensions in Explain by
- Use plain field names, since they become level titles
- Lock the first levels when every reader needs them
- Give the visual the full width of the page
- Place a slicer nearby so readers can change the period
Power BI decomposition tree: common questions
Which field can go into Analyze in a Power BI decomposition tree?
Microsoft states that the Analyze well takes the metric you want to study, and it must be a measure or an aggregate. A plain text column will not work there. The Explain by well then holds one or more dimensions, which readers use to split the total.
Why did a branch disappear after I filtered the page?
Because cross-filtering removed its data. In Microsoft's example, filtering by one publisher leaves a platform blank, so that node and its path drop out of view. However, the existing levels stay pinned on the tree. When you select another node, the tree expands to those levels again.
Can readers change the levels of a decomposition tree in Power BI?
Yes, unless the author locked them. Readers can remove a level, pick another field or add new levels after the locked ones. A locked level stays as it is, but readers can still select other nodes inside it. So an author can guide the start and leave the rest open.
Why does a level show only ten values?
Microsoft documents that the tree cuts each level to its top values, and the current number per level is 10. So a field with many distinct values will not show all of them. If readers need a specific value, add a slicer or a filter for that field.
Can Power BI Agency build a root-cause page for our team?
Yes. One of our consultants agrees the measure and the dimensions that matter with your team, checks the source connection and then builds the page. We work online inside your own Microsoft tenant and show progress in weekly demos. Power BI Agency is a division of Sataware Technologies, independent of Microsoft.
Let's build your first dashboard together.
Tell us what you report on today. A consultant replies within 2 hours with next steps.
Trusted partners
Microsoft Solutions Partner · AWS Partner
Cloud & AI Platforms
Azure SQL, Microsoft Fabric and the data pipelines behind your reports.
AI Business Solutions
Power BI reporting on your Dynamics 365, Business Central and Power Platform data.
Security
Row-level security and Entra ID sign-in on every report and workspace.
Amazon Web Services
Power BI on your AWS data, with connectors for Redshift, Athena and S3.