Changes to data, reporting, and analytics services vary significantly in complexity and risk. Some are simple service improvements. Others may affect trusted figures, operational decisions, or access to sensitive information.
To support timely delivery while maintaining confidence, changes should follow a route proportionate to their impact.
This matrix helps classify requests consistently and select the right level of control.
Core Principle
Not every change needs the same process.
Low-risk improvements should move quickly. Higher-risk changes should receive stronger review, validation, and communication.
Change Levels Matrix
| Level | Category | Typical Risk | Examples | Typical Route |
|---|---|---|---|---|
| Level 0 | Cosmetic / Safe | Very Low | Labels, formatting, descriptions, layout changes | Fast-track |
| Level 1 | Controlled Safe | Low | Source replacement with same meaning, performance tuning, technical tidy-up | Peer review + validation |
| Level 2 | Business Logic | Medium | KPI rule changes, new calculations, status mapping, filters affecting totals | Impact review + stakeholder approval |
| Level 3 | Structural / Sensitive | High | Removed fields, join/grain changes, security changes, major model refactor | Formal release control |
| Level 4 | Strategic / Platform | Very High | ERP transition, warehouse redesign, platform migration, operating model change | Programme governance |
Level Guidance
Level 0 — Cosmetic / Safe
Improves presentation or usability without changing meaning.
Examples:
- Correcting labels or spelling
- Reordering columns
- Improving navigation
- Visual formatting changes
- Help text updates
Expected outcome: Faster user experience with no data impact.
Level 1 — Controlled Safe
Technical changes where outputs are expected to remain materially the same.
Examples:
- SQL optimisation
- Refactoring code for maintainability
- Source table renamed with identical mapping
- Dataflow improvement with same result set
Expected outcome: Better service quality with low business risk.
Level 2 — Business Logic
Changes that may alter numbers, definitions, or interpretation.
Examples:
- Margin calculation changes
- Revised order intake logic
- Status lifecycle reclassification
- Filter rules affecting totals
- New KPI definitions
Expected outcome: Improved business logic with managed stakeholder awareness.
Level 3 — Structural / Sensitive
Changes that may break dependencies or affect security.
Examples:
- Column removal
- Primary key redesign
- Row-level security changes
- Join grain changes
- Major semantic model restructure
Expected outcome: Controlled release with stronger validation and communication.
Level 4 — Strategic / Platform
Large-scale change involving programmes, platforms, or enterprise transformation.
Examples:
- ERP migration
- New reporting platform
- Data warehouse re-architecture
- Regional operating model consolidation
Expected outcome: Managed transition with executive visibility.
Quick Assessment Questions
When assessing a change, ask:
- Does it change trusted totals?
- Does it change business meaning?
- Does it remove or rename relied-upon fields?
- Does it affect security or permissions?
- Does it impact multiple reports or teams?
- Is rollback difficult?
If the answer is yes to any material question, the change may require a higher level.
Important Rule
Where multiple categories apply, use the highest relevant level.
For example:
- A formatting improvement that also changes security is not Level 0.
- A performance improvement that changes totals is not Level 1.
Why This Matters
Using a common matrix helps us:
- Prioritise effort effectively
- Reduce unnecessary delay
- Focus governance where risk exists
- Improve consistency across teams
- Communicate clearly with stakeholders
- Maintain trust in reporting outputs
Related Guidance
See also:
- Safe Data Explained
- How to Raise a Change
- Recent Changes
- Breaking Changes Register
- Release Calendar
Final Note
Good change control is not about slowing progress. It is about applying the right level of control to the right type of change.