SQL and Python
Write or refine logic directly when you need complete control.
Data Studio
Write the logic behind your numbers.
SQL, Python, notebooks, or the Agent, on the same prepared and access-controlled data. Test the logic, then promote it to a scheduled model with visible run history.
Inspectable logicScheduled modelsBQML support
Investigate with Agent
Same data. Logic shown.
Question
Which channels are growing revenue without increasing CAC?
SELECT
channel, revenue, cac,
revenue_growth_30d
FROM growth_performance
WHERE data_quality = 'passing'
Preview 100 rows
Data definitions applied
Work your way
Build with SQL, Python, notebooks, or AI.
Choose the interface that fits the question and your level of control. Every method uses the same data and can produce the same saved result.
Notebooks
Explore the same prepared, access-controlled data without exporting it to another tool.
Agent-assisted analysis
Ask a question in plain language and inspect the working behind the answer.
Scheduled models
Promote tested logic into recurring work with visible run history.
Machine learning
Build predictions and classifications, including supported BQML workflows.
Scenario planning
Compare spend and performance assumptions before committing budget or making a change.
From model to action
Save the result as a field, a table, or a data sync.
Write a result back as a calculated property, keep it as a reusable table, or send an approved output to a supported destination.
Customer properties
Scores, tiers, predictions, and calculated values.
Events and sessions
Channel groups, classifications, and attribution credit.
Accounts and tables
Reusable outputs for B2B reporting and shared analysis.
Destinations
Approved groups and properties sent to supported tools.
Starting points
Use prepared attribution, Customer 360, or retention models.
Start from a common marketing model, then inspect and adapt its assumptions to your data rather than beginning from a blank page.
Multi-touch attribution
Compare methods and lookback windows against your own first-party data.
Predicted LTV
Estimate future customer value and make it available for reporting and groups.
Churn risk
Identify customers whose recent behavior resembles historical churn patterns.
Purchase propensity
Score purchase likelihood and create a reusable high-intent customer group.
FAQ