Microsoft Fabric promises an end-to-end analytics platform, but when it comes to testing, the reality is far less clear-cut.

In this session, I’ll share what testing Fabric actually looked like on a large government transformation programme, where data pipelines, Power BI, CI/CD, and automation all collided under real delivery pressure. This isn’t a “best practice” talk—it’s an honest account of what worked, what broke, and what we had to rethink along the way.

We’ll start by setting the context: what Fabric changes about testing, where traditional approaches fall down, and the new risks that appear when you move from APIs and UIs to analytics, pipelines, and reports. From there, I’ll walk through concrete examples from real code, pipelines, and Power BI artefacts—covering data validation, flaky automation, CI/CD integration, and how we decided what not to automate.

The session closes with a worked example and a clear set of takeaways that teams can apply immediately.

You’ll leave with:

Practical approaches to testing Fabric pipelines and Power BI outputs without chasing false confidence

Ways to define meaningful test oracles for data quality and analytics

Examples of automation patterns that helped—and anti-patterns that slowed teams down

Ideas for structuring CI/CD feedback in fast-moving data platforms

Session structure (flexible: 40–60 minutes)

Core talk (40 minutes):

10 mins – Context: Fabric, analytics testing, and real-world constraints

20 mins – Lessons learned: pipelines, visuals, automation, and CI/CD

10 mins – Worked example and concrete takeaways