Project · Advanced · about 20 hours
Secure self-serve sales model for 27 regional managers
Turn a report into a governed, documented semantic model with row-level security, refresh and a release process.
Skills:Data modelling, Power BI, Stakeholder management
Part of: Data Analyst
Data and licences
- Olist Brazilian E-Commerce Public Dataset (Kaggle) ↗CC BY-NC-SA 4.0 (non-commercial use, share alike)
Brief
The marketplace has managers for each of Brazil's 26 states and the Federal District, 27 in all. Each should explore sales for their own state only, build their own reports on a shared model, and trust that numbers match the head office's. You own the model.
What to build
Run a requirements conversation (with a friend playing the head of sales) and write a one-page brief: users, questions, security rules, refresh needs.
Build the model from the builder-level star schema, with a measures table, display folders and descriptions on every measure.
Implement dynamic RLS with a user-to-state mapping table, and test with "View as" for three users, including one with two states and one with none.
Set up development and production workspaces (or describe them if you have no licence), a refresh schedule, and a change log.
Deliverables
.pbixwith the model,brief.md,rls-tests.mdwith screenshots,RELEASE.md
How it is graded
The reviewer impersonates your test users and checks what they can see. Security mistakes cap the score. Then clarity of documentation and whether the model answers the brief.
Stretch
Add calculation groups for time intelligence and compare with separate measures.
Deliverables
- A Power BI semantic model with documented measures
- Dynamic row-level security by customer state, tested with at least three users
- A requirements brief agreed with a (role-played) stakeholder
- [object Object]
How a reviewer grades it
| Criterion | Weight | What good looks like |
|---|---|---|
| Requirements captured and reflected in the model | 20% | |
| RLS correct and tested, including edge cases | 30% | |
| Measures documented and consistent | 25% | |
| Release and refresh process is realistic | 25% |