Azure Data Platform Modernization Playbook

Tell us what you're building

We use these answers to surface the prompts, skills, and MCP configs that fit your stack — and to substitute stack values like {{database}} into the prompts you copy. Content (PRDs, code, etc.) stays in your repo. Everything you enter here is stored in your browser — nothing is sent to a server.

Steps you'll go through

1. 01

   Discovery and Source Inventory

   4-6 weeks

   Profile every major source system, document quality issues, identify downstream consumers, and capture stakeholder pain points. This is a stakeholder conversation, not just a tech audit.

2. 02

   Target Architecture Design

   3-4 weeks

   Make the Fabric vs Synapse vs Hybrid call. Design the lakehouse (medallion). Decide storage layout, compute strategy, orchestration, consumption layer, and governance — then capture as ADRs.

3. 03

   Pipeline Framework Design

   4-6 weeks

   Build the metadata-driven framework before building any specific pipelines. Multiplier work — done right, every subsequent pipeline is faster.

4. 04

   PySpark Transformation Framework

   3-4 weeks

   Establish PySpark standards before writing many notebooks. Shared utility library, Bronze / Silver / Gold templates, performance patterns, testing approach.

5. 05

   Data Quality Framework

   3-4 weeks

   Build the DQ framework before pipelines ship to production. Schema, freshness, volume, nulls, duplicates, ranges, referential integrity, business invariants.

6. 06

   First Domain (Vertical Slice)

   8-12 weeks

   Migrate / build one complete business domain end-to-end (Customer or Sales is typical). Bronze + Silver + Gold + DQ + a Power BI semantic model. Validates everything from Phases 3-5 against a real use case.

7. 07

   Continue Expanding Domains

   1-3 quarters

   Build out remaining domains, refining patterns. Common slicing: by business domain, by source system, by criticality, or by region.

8. 08

   Migration of Consumption (BI, ML, Downstream Apps)

   1-2 quarters (often parallel with Phase 7)

   Switch consumers from legacy data sources to the new platform. Power BI, Tableau, ML pipelines, custom apps, Excel queries — each gets a different migration path.

9. 09

   Decommission Legacy + Optimize

   4-8 weeks for decommission, ongoing for optimization

   Turn off old systems. Run cost optimization (typically 30%+ savings opportunity). Tune performance. Mature governance.