Data modeling in AI/BI dashboards

Every AI/BI dashboard visualization reads from a dataset. On top of datasets, AI/BI dashboards give you several ways to shape and reuse your data without pre-joining tables in SQL or duplicating logic across datasets. Each option fits a different need, from lightweight per-dataset calculations to multi-fact semantic models, so you can choose the one that matches your analysis.

You can define modeling logic directly in a dashboard when it doesn't yet belong in the governed Unity Catalog layer, then promote it to Unity Catalog when it's ready for broader use. For metrics that must be governed and shared across dashboards, Genie Spaces, or other tools, use Unity Catalog metric views. See Unity Catalog metric views.

Datasets

Datasets are the queries or tables that supply data to a dashboard. Every other modeling option builds on top of a dataset. See Create and manage dashboard datasets.

Custom calculations

Custom calculations let you define new measures and dimensions computed from an existing dataset without changing the source SQL. Use them for lightweight metrics and transformations, such as a profit margin measure or a categorized dimension, that are scoped to a single dataset. See What are custom calculations?.

Local metric views

Local metric views let you define dimensions, measures, and join relationships in the dashboard using a visual interface, while keeping the semantic logic contained within a single dashboard. They retain the benefits of semantic objects, such as accurate aggregations regardless of grouping, and you can promote them to Unity Catalog when your metrics are ready for broader use. Local metric views are in Public Preview. See Local metric views.

Dashboard relationships

Dashboard relationships let you define a semantic model that spans multiple fact and dimension tables by specifying how datasets relate to one another. The query engine resolves joins at runtime, so you can slice fields and measures across any connected dataset and build multi-fact, multi-grain analysis in a single dashboard. Dashboard relationships are in Public Preview. See Dashboard relationships.

Choose the right approach

The dashboard-scoped modeling options solve different problems. Use the following table to decide which one fits your analysis.

Approach	What it provides	Best fit
Custom calculations	New measures and dimensions computed from a single dataset, without changing the source SQL	Lightweight metrics and transformations scoped to one dataset
Local metric views	Dimensions, measures, and joins defined in the dashboard at a fixed grain, promotable to Unity Catalog	Reusable semantic logic that you might later promote to a governed Unity Catalog metric view
Dashboard relationships	A traversable join graph across multiple datasets, with reusable cross-dataset measures	Multi-fact, multi-grain analysis that spans several datasets in one dashboard

Fixed grain compared to dynamic grain

Granularity affects which fields you can select and the grain at which they're expressed. Metric views lock a table at a fixed granularity, for example customer-level granularity, whereas dashboard relationships are dynamic, based on the fields used in the dashboard, for example customer, order, or shipment-level granularity.

Both can model joins, but they solve different problems. Dashboard relationships are more flexible: they let you mix and match fields and measures from any table across your entire semantic graph. A dashboard relationship graph can include metric views as nodes, with relationships as the edges that connect them. The metric view handles the single-grain logic, and relationships handle the multi-fact layer.

Dashboard-scoped modeling compared to Unity Catalog

Local metric views and dashboard relationships are scoped to a single dashboard and do not create a Unity Catalog object. Use dashboard-scoped modeling for prototyping, dashboard-specific analysis, or when you want to iterate before promoting logic to Unity Catalog.

Use Unity Catalog metric views when metrics must be governed, queryable directly with SQL from any SQL client, and reusable across dashboards, Genie Spaces, and other tools. When a local metric view is ready for broader use, you can export it to Unity Catalog. See Export to a Unity Catalog metric view and Unity Catalog metric views.

In this section

Page	Description
Datasets	Create and manage the datasets that supply data to a dashboard.
Custom calculations	Define dynamic measures and dimensions without changing dataset queries.
Local metric views	Define dimensions, measures, and joins in the dashboard, then promote them to Unity Catalog.
Dashboard relationships	Model joins across multiple datasets to build multi-fact semantic models.