What is a semantic model (Power BI dataset)?
A semantic model, formerly called a Power BI dataset, is the semantic data layer in Power BI that bundles tables, relationships, metrics and permissions. It forms the reusable foundation on which multiple reports build with consistent definitions.
Also known as: Power BI dataset · data model · Power BI semantic model · tabular model
Where a semantic model is used
The semantic model is the layer between the raw data and the reports. It contains the loaded tables, their relationships, the metrics defined with DAX and security rules such as row-level security. Reports do not access source data directly but this model, which keeps metrics and definitions consistent across all reports.
A centrally maintained semantic model can be reused by many reports and users. This avoids the same metric being calculated differently in ten reports and makes the model a shared source of truth within Power BI.
Typical use cases
A semantic model pays off wherever several reports or teams rely on consistent metrics.
- A central model as the basis for many reports and dashboards
- Consistent metric definitions across departments
- Centrally maintained permissions, for example via row-level security
- Reuse instead of redundant models in every single report
How it relates & how smiit uses it
The semantic model is the model layer in Power BI, not the data warehouse below it and not the report above it. Power Query fills it, DAX defines its metrics, and row-level security governs visibility. With large data volumes, an upstream data warehouse or lakehouse is recommended as the model's data source. In the dy Project AG data platform, the semantic model builds on the gold layer and provides validated metrics. smiit builds semantic models so that they stay performant, reusable and maintainable in everyday work.
Common mistakes & misconceptions
- A semantic model is not just a copy of data; it holds relationships, hierarchies and measures that turn raw data into a business-friendly layer.
- Many still know it as a dataset and treat the two as different. In Power BI the former dataset was renamed semantic model and refers to the same concept.
- A common error is to think every report needs its own model. A well-maintained, shared semantic model avoids redundancy and conflicting metrics.
Frequently asked questions
Why is the Power BI dataset now called a semantic model?
Microsoft renamed the term dataset to semantic model to clarify that it is a semantic data layer with relationships, metrics and permissions, not just a plain data table.
Can several reports use the same semantic model?
Yes. That is precisely the benefit: one centrally maintained semantic model supplies many reports with the same metrics and definitions, creating consistency and avoiding duplicate work.
What is the difference between a semantic model and a data warehouse?
The data warehouse stores the prepared data, while the semantic model is the layer above it in Power BI with relationships, metrics and permissions. With large data volumes, the warehouse serves as the model's data source.
How do you keep a semantic model maintainable in the long run?
Clearly named metrics, a well-thought-out data model with clean relationships and avoiding redundant calculations all help. When logic and definitions are maintained centrally, changes stay traceable and automatically affect every report built on the model.
What does smiit pay attention to when building a semantic model?
smiit builds semantic models so that they stay performant, reusable and maintainable in everyday work, and with large data volumes builds them on a validated data foundation such as the gold layer of a data platform.
Related terms
Sources & further reading
Want to put this topic to work in your company?
Updated · Back to the glossary