Models

Models let you transform and shape your data in BigQuery before syncing it to destinations. Use models to create custom metrics, build predictive features, or define audience segments for ad platforms and CRMs.

Model Types

Vendo supports three types of models:

SQL Models

Write SQL queries against your BigQuery tables to create custom views, aggregations, and transformations.

• Define a query, validate it against your schema, and schedule it to run
• Output types: events, user profiles, group profiles, lookup tables, or custom
• Supports incremental syncing via primary key and incremental column

Learn more about SQL Models →

Python Models

Use pre-built templates or write custom Python code for advanced transformations like ML predictions, customer segmentation, and statistical analysis.

• Template categories: recommendations, segmentation, prediction
• Map BigQuery tables as inputs, configure output table and write mode
• Schedule independently or trigger on source data arrival

Learn more about Python Models →

Audiences

Build rule-based user segments with a visual editor. Audiences generate targeted user lists for ad platforms (Customer Match, Custom Audiences) and messaging tools.

• Define rules with fields, operators, and values
• Map identity fields (email, phone) for PII hashing
• Sync audiences directly to Google Ads, Meta Ads, OneSignal, Klaviyo, and more

Learn more about Audiences →

How Models Fit in the Pipeline

Sources → BigQuery → Models → Destinations

1. Sources import raw data into your BigQuery dataset
2. Models transform, aggregate, or segment that data
3. Destinations receive the modeled output — events, profiles, or audience lists

Models can also chain together: a Python model can trigger after a source sync completes, and a downstream model can trigger after an upstream model finishes.

Syncing Models to Destinations

Any model can be synced to a compatible destination. From the model detail page, click Sync to Destination to create a connection that exports the model’s output.

Compatibility depends on the model’s data type:

Related

• Attribution Models — Configure how conversion credit is allocated across touchpoints
• Data Science Agent → Building Models — Create models with AI assistance