Data Studio
Turn source data into reusable datasets with SQL, Python, notebooks, grouping rules, or supported models, then validate the output.
Last reviewed September 15, 2026
Data Studio models turn raw source data into reusable datasets and activation-ready outputs. In the app, open Studio from the top menu.
Data Studio gives you multiple approaches in one layer:
| Type | Best for |
|---|---|
| SQL Models | Views, aggregations, joins, and custom metrics |
| Python Models | Advanced logic, ML preprocessing, and statistical analysis |
| Grouping Rules | Rule-based segmentation and categorization |
| Notebooks | Interactive exploration with browser-side Python |
| BQML Models | Predictive ML trained inside BigQuery |
| Attribution Models | Multi-touch attribution |
| Audiences | Audience lists for ad platforms |
All model outputs can attach to output targets: user, event, session, or account properties.
Create a model
- Open Studio.
- Click New Data Model.
- Choose an approach: SQL, Python, Notebook, Grouping Rules, or BigQuery ML.
- Configure inputs, logic, and output target.
- Set the run policy, or run the model on demand.
You can also ask the Agent to create a model in a conversation.
Calculated properties
Vendo can make calculated properties in 2 ways:
- Rule-based: SQL CASE expressions that Vendo evaluates at query time.
- Precomputed: scheduled materialization into user, event, session, or account properties.
Both types appear in the Data Catalog next to source properties. See Customer 360 for the built-in property templates.
Pipeline integration
Data Studio models connect to the orchestration graph. When an upstream source sync completes, downstream models can start automatically. This makes chains like this possible:
Source Import --> Property Compute --> BQML Retrain --> Destination SyncRelated
- Output Targets: attach outputs to catalog properties
- Customer 360: identity and calculated properties
- Agent: build models in a conversation
- Orchestration: operational graph