Data Model Optimization
Data Model Optimization helps you detect and eliminate duplicate or redundant logic across your data models. When multiple pipelines perform similar transformations, Zingle identifies merge opportunities and lets you consolidate them via pull request.
The optimization dashboard
Navigate to Data Model Optimization in the sidebar. The main view shows:
- Total estimated savings across all identified opportunities
- Number of affected models that could be consolidated
- Duplicate logic detected — count of redundant transformation patterns
Click into any opportunity to see details.
Reviewing an optimization opportunity
Each opportunity card shows:
| Field | Description |
|---|---|
| Opportunity ID | Unique identifier |
| Affected models | List of models with overlapping logic |
| Duplicate pattern | Description of the redundant transformation |
| Estimated savings | Cost reduction from merging |
| Suggested action | Merge recommendation |
Merging models
Open the merge view
Click into an optimization opportunity and select Merge. This opens a dedicated merge interface showing:
- Before lineage — current state with redundant models
- After lineage — proposed consolidated state
- Data comparison — side-by-side data preview
- SQL comparison — diff of the transformation logic
Review the proposed changes
Examine:
- The consolidated SQL to ensure correctness
- The updated lineage to verify no downstream dependencies break
- Data samples to confirm output equivalence
Edit if needed
If the suggested merge isn't quite right, edit the suggestion directly in the interface before committing.
Create the merge PR
Select a GitHub repository and click Create PR. Zingle:
- Creates a branch for the merge
- Commits the consolidated model and removes redundant files
- Opens a PR with full context on what was merged and why
When to use this
- After a period of rapid model creation — teams often build similar models independently
- During quarterly reviews — identify consolidation opportunities to reduce warehouse spend
- Before scaling — clean up technical debt before adding more complexity