User documentation

Audit your Power BI models. Identify. Prioritize. Optimize.

PBI Audit deeply analyzes your Power BI semantic models and provides a global score along with precise recommendations to improve performance, quality and governance.

Get started
Modeling
Detection of incorrect schemas, problematic relationships, isolated tables and more.
View rules →
DAX
Static analysis of your DAX measures: anti-performance patterns, bad practices, deep dependencies.
View rules →
VertiPaq
Storage engine analysis: cardinality, compression, unused columns, partitioning.
View rules →
Performance
Actual execution of your DAX measures to measure FE, SE and CPU time.
View rules →
Governance
Missing descriptions, naming conventions, RLS roles, KPI measure formats.
View rules →
Power Query
Detection of sub-optimal M transformations that slow down loads and cause Capacity Unit overconsumption on Fabric.
View rules →

What is PBI Audit?

PBI Audit is a Windows tool dedicated to Power BI developers and tech leads. It connects directly to your semantic models, whether from Power BI Desktop or from the Power BI Service via an XMLA endpoint, and performs a full audit in seconds.

The result is a global score out of 100, broken down by domain, accompanied by a list of findings ranked by severity. Each finding precisely indicates the concerned object (table, column, measure, M query) and the associated recommendation.

The goal: give you a clear and actionable view of your model's state, without having to manually browse through dozens of best practices.

Who is this tool for?

Power BI developers
Validate model quality before delivery, identify slow DAX measures, spot anti-performance patterns you might have missed during a manual review.
BI tech leads & architects
Quickly diagnose why a model over-consumes Fabric Capacity Units, why reports load slowly or why the SE engine is saturated.

Score at a glance

The global score is the weighted average of the scores for each analyzed domain. A non-analyzed domain (for example, Performance if trace rights are not available) is excluded from the calculation and its weight redistributed.

≥ 80
Good
60 – 79
To improve
< 60
Critical
Each rule applies a capped penalty according to its severity: maximum 15 points for a high rule, 8 points for medium, 4 points for low. A single rule can never drop the score to zero.