If your Power BI estate feels like a share-house fridge—everyone “borrows”, no one labels, and something questionable is growing at the back—you’re not alone.

A new metric lands, a team spins up a report, another team copies it, and the tenant sprouts fresh vines. FOMO creeps in: what’s out there, who’s using it, and are we still safe - especially where patient data might surface?

You know that sinking feeling when someone asks, “Who’s looking at that report with patient data?” and you’re not 100% sure?

That’s admin FOMO — Fear Of Missing Oversight.

FUAM (Fabric Unified Admin Monitoring) is the comprehensive community accelerator that turns the lights on. It givesyou a living, tenant-wide view of what exists, who’s using it, what’s breaking,and where the risk is - so you can move from guesswork to governance.

What FUAM Actually Does

• Builds your governance catalogue: a reliable inventory of workspaces, reports, semantic models, dataflows, pipelines, notebooks and refreshes—centralised and queryable.
• Surfaces real usage: who viewed/exported/printed what, when, and from where (apps vs workspaces vs embedded), so you can separate who can from who did.
• Shows capacity & reliability: refresh trends, failures, query pressure and noisy neighbours - ideal for ops stand-ups and Fabric capacity right-sizing.
• Highlights security signals: owners, admins, endorsement status, “publish to web” flags, orphaned workspaces/items, and access-grant events for audit trails.
• Enables ad-hoc questions: this one is one of my favourites - I use the SQL end point for many queries I have because the data lands in a Lakehouse (yes it uses Fabrics heart and soul in the architecture) with a ready semantic model, custodians can run their own queries without wrangling raw log

‍

Why this matters for Healthcare, Governance and Privacy

FUAM is the start of your report governance catalogue and the home for any ad-hoc queries your Power BI custodians need to pull critical governance metadata. From there, you can tag reports that might surface patient data and carry sensitivity/privacy requirements. That gives you a systematic audit path through a jungle-like environment that’s grown organically over years as teams and data needs evolved.

How we’ll use FUAM for sensitivity oversight (no drama)

1. Stand up FUAM in a Premium/Fabric workspace and run the first load.

‍‍‍2. Add a lightweight “Sensitivity Register” table in the FUAM Lakehouse.
    (keys: WorkspaceId, ReportId, DatasetId; fields: Sensitivity Tag, Data Owner, Review Date, Notes)

3. Triage list: use FUAM’s usage and export signals to prioritise which reports get reviewed/tagged first (e.g., high-usage + exports = top of the queue).

4. Owner attestation: owners confirm whether a report can surface patient-identifiable data; tag it (e.g., PII/PHI, internal-only, external-safe.

5. Ongoing hygiene: schedule monthly reviews, track remediations (add RLS, remove publish-to-web, restrict export), and monitor drift via FUAM.

What FUAM won’t do (and that’s okay for now)

It won’t read your column values or auto-decide whether something is PII/PHI.

Near term solution: keep tagging data-owner–led, backed by FUAM’s usage, export and ownership signals.

Longer term solution: consider a local, self-hosted AI agent to categorise columns (categorisation is a key task AI makes light work of), suggest labels and nudge data owners and custodians, taking most of the manual legwork out of the process.

‍

Longer Term Solution - Brainstorm

Below is a diagram depicting a possible solution bringing together both FUAM and AI to help with Sensitivity and Governance Reporting:

‍

Swim lanes (top → bottom)

Governance – Data owners and custodian (human loop) steps: review, validation, and tasking.

Fabric – Unified Data platform: ingestion, Lakehouse, Delta tables, and Reporting.

Azure – Cloud services that do the heavy lifting (AI + broker).

Security – Guard rails used end-to-end (identity, keys, logging, private access).

Components and Flows Explained

Ingest (Fabric)
Data and FUAM metadata are ingested from your Fabric tenant into the platform on a schedule/incremental pattern. This seeds the pipeline with ownership, activity, usage and schema signals.

FUAM Lakehouse (Fabric)
Landing/curation zone for FUAM datasets. Normalises and readies metadata for downstream processing, and exposes conformed tables to notebooks and services.

‍OpenAI Open Source Model - self-hosted (Azure Container)
Private, self-hosted model classifies assets using FUAM metadata (and policy prompts) to propose sensitivity classes and candidate tags. No public egress. Open models by OpenAI | OpenAI‍‍

Heuristic Notebook (Fabric)
Deterministic rules complement the AI: pattern/keyword checks (e.g., DOB, URN, Medicare) generate draft tags with rationale and confidence. Writes results to staging.

‍Human-in-the-loop– Data Custodian (Governance)
Custodians triage suggestions in context—confirm/override/add notes. This step captures accountable decisions and creates an auditable trail.

‍Azure Function – Broker (Azure)
Orchestrates prompts and categorisation across AI + rules, manages throughput/retries, and persists outcomes back to Fabric. Acts as the glue between services and the app.‍‍

Sensitivity Register App & Orchestration Pipeline(Governance ⇄ Fabric)
The Power Apps front-end records custodian validations; the pipeline writes these decisions into the Sensitivity Register (Delta table) as the single source of truth. A single source of truth is critical for reporting and auditing!

Governance Reporting (Fabric)
Power BI models the Sensitivity Register and FUAM signals to deliver coverage, risk, exceptions, and owner views. Feeds Data Owners & Custodians for action.

‍Notifications & Tasks (Governance)
Power Automate raises tasks with due dates, reminders, and escalations (e.g.,via Teams/Outlook) to drive closure of outstanding triage and remediation actions. 

The Punchline

For hospitals and health services—FUAM is the meta data machete-and-map. It builds a living reporting catalogue across your Fabric tenant, shows real usage, surfaces owners and risks early, and sets sensible guardrails without slowing Power BI/Fabric delivery—so governance stays tightwhile improvement work keeps humming.

It gives you the operational backbone for a reporting governance catalogue. Health Information Managers, Privacy Officers and Data Custodians can answer the hard questions quickly, tag sensitive content (think URNs, Patient Names, DOBs, Medicare numbers), and keep the ever-growing data estate tidy. Incident response, audits and accreditation prep get faster and less stressful.

‍This would really bolster your health service data governance and privacy capability come accreditation!

Bottom line

Next time the exec team, auditors or the board member asks “who’s got Power BI access to what—and why?”, you’ll have the evidence before morning's exec huddle.

If you’re a public or private hospital, mental health, community health service or aged care using FUAM—or keen to roll it out—yell out and we’ll help shape a practical, right-sized approach for your organisation. 

‍

Credit and Recognition

FUAM official community GitHub repo: fabric-toolbox/monitoring/fabric-unified-admin-monitoringat main · microsoft/fabric-toolbox · GitHub

The FUAM solution accelerator is not an official Microsoft product! It is a solution accelerator, which can help you implement a monitoring solution within Fabric.

‍

Written By: Luke Garton - Principal Consultant, proofed by AI.

#FUAM #Fabric #PowerBI#Healthcare #DataGovernance #FOMO #Sensitivity #GovernanceReporting

‍