What it is

What is the Fabric Accelerator?

The plain answer, the mechanism behind it, and every feature it ships with, explained in terms of what it does for you, not just what it's built from.

In one sentence How it's built Every feature Connectors Governance

A metadata-driven Microsoft Fabric data platform: a pre-built bronze, silver and gold pipeline, with governance, monitoring and reporting-ready models included.

Configured to your source systems. Not commissioned and built from a blank workspace.

Not a template you build on top of. Not a discovery phase with a platform at the end. A working product, deployed into your own Fabric tenant and pointed at your systems.

It already exists

The architecture, the ingestion framework, the masking and the history logic were built and proven before your engagement started, on someone else's timeline, not yours.

It's configured, not coded

Connecting a source, changing a load type, adding a batch schedule: all of it happens through configuration, the same framework every time.

It's yours to run

Deployed inside your own Microsoft tenant, on your own data. Documented and teachable, so your team can operate it without us.

The mechanism

How it's built.

Every layer below is prebuilt. Source data moves through the same four stages whatever system it came from, so a new connector extends the framework instead of becoming a one-off pipeline.

Your sources

  • Finance system
  • CRM
  • Operations
  • Files and APIs
→

Bronze · raw

Landed exactly as received, with full history retained. Nothing edited, so you can always prove where a number came from.

→

Silver · conformed

Cleaned, typed and joined. One customer record, one product list, one version of each entity across every system.

→

Gold · business

Modelled to how you actually report. Agreed definitions, so revenue means one thing in every dashboard.

→

Consume

  • Power BI
  • Fabric AI features
  • Excel
  • Downstream apps
Built in across every layer PII maskingSlowly changing dimensionsMetadata-driven ingestionCI/CD across environmentsRefresh monitoring and alerting

New source added by configuration, not code. Same framework, same governance, no new build.

AI is only as good as what it's pointed at.

Copilot and Fabric's AI features are genuinely useful, but they inherit whatever they're given. Point them at four systems that disagree with each other and you get confident answers that happen to be wrong, produced faster than anyone can check them.

The gold layer is what makes the difference: agreed definitions, traceable lineage, one version of each figure. That isn't preparation for AI, it's the same thing that makes your Monday reporting trustworthy.

Every feature, plainly

What it does, and why it's there.

Ten things the platform does out of the box. Each one exists because it's the thing that goes wrong when a data platform is built in a hurry.

Connects to what you already run

Metadata-driven connectors for SQL Server, Azure SQL, Fabric Warehouse, CSV/Excel, SharePoint Lists, REST APIs and Dataverse. Adding a source is a configuration form, not a new pipeline.

Why it matters

Most businesses run more source systems than they'd like to admit. When you acquire a business or swap a system next year, connecting it is a form to fill in, not a project to commission.

A raw layer nobody can argue with

Bronze/silver/gold architecture. Source data lands untouched with full history, gets conformed into one version of each entity, then reshaped into business-ready gold tables.

Why it matters

When someone asks where a number came from, you point at the layer instead of asking the one person who remembers. Nothing is edited in place, so every figure traces back to its source row.

Personal data is masked by default

Dynamic data masking applied at the point of query, enforced in the Fabric SQL endpoint itself. Unmasked access is granted through named security groups sourced from your own Entra ID tenant.

Why it matters

Governance in a policy document gets skipped under deadline pressure. Governance in the database engine doesn't have that failure mode.

History that doesn't disappear

Slowly Changing Dimension (Type 2) tracking on silver entities. When an address, a price or a role changes, the old version is kept and dated, not overwritten.

Why it matters

"What did we believe was true on the day that order was placed" is a question finance and audit ask more often than people expect. This platform can answer it for any point in the past, not just today.

One customer record, not four that disagree

Automatic surrogate key generation and foreign key resolution, so the same customer, product or order is a single joined-up entity in gold, however many systems it came from.

Why it matters

This is the actual work behind "one version of the truth": a specific technical step that happens automatically instead of being redone by hand in every report.

Runs reliably, and tells you when it doesn't

Batches run as ordered execution stages with partial-failure retry: if three sources succeed and one fails, only the failed one reruns. Every run is logged with its cause.

Why it matters

A pipeline that reruns everything on any failure wastes time; one that fails silently is worse. This does neither: it retries precisely and leaves an audit trail an operator can read.

A health dashboard, not a support inbox

Pre-computed health statistics (typical duration, row counts, consecutive failures) compared against each run in real time. Email alerts fire automatically on pipeline failure.

Why it matters

You hear about a problem from the platform, before the FD asks why Monday's numbers look wrong. Not from one person quietly watching an inbox.

Reports ready to build on, not a data dump

Gold-layer tables ship with an auto-generated semantic model (TMDL) for Power BI: relationships, hierarchies and measures already defined.

Why it matters

The gap between "the data landed" and "someone can build a trustworthy report from it" is usually where a data project loses months. This closes that gap as part of the platform.

Access tied to identity you already manage

Sign-in and role assignment run through your own Microsoft Entra ID tenant: Consultant, Admin, Editor, Viewer, and a restrictive default for anyone unrecognised.

Why it matters

One less identity system to secure and audit. Someone leaving the business is handled by your existing offboarding process, with nothing platform-specific to remember to switch off.

The same platform, whatever size you grow to

One framework, deployed per client, with core updates flowing down from a shared codebase. Runs on Fabric's smallest capacity SKU and its largest without a redesign.

Why it matters

Growth becomes a capacity setting, not a re-platforming project. Because it's documented and configuration-driven, your own team can run it after we leave.

Runs on your stack

Connects to what you already run.

Every connector lands through the same ingestion framework, so a new source is a configuration exercise, not a new build.

  • Dynamics 365
  • Business Central
  • Dynamics F&O
  • Dataverse
  • SQL Server
  • SharePoint Lists
  • Excel
  • CSV and Text Files
  • JSON
  • Xero API
  • REST APIs

Not on the list? New connectors are added to the framework, not bolted on as one-off code. Ask on the call.

Governed by design

An advocate for governance.

Masking, access control, audit and data-quality checks sit at the centre of every layer, not layered on as an afterthought. Nine governance mechanisms, built in from the first deployment.

Fabric
Accelerator
Masking Irreversible PII masking in the silver transform chain (salted hash, nullify, redact, year-only date, or partial reveal) before silver is ever written.
History Full slowly-changing history retained through every layer, so any figure can be traced back to the version of the record it came from.
Monitoring Health statistics and anomaly detection on every run: duration, row counts and consecutive failures compared against a documented baseline.
CI/CD Every environment deployed the same way, every time, so a change is reviewed and repeatable rather than a manual edit in production.
Ingestion Metadata-driven ingestion: a new source is a configuration row, not a new pipeline to write and review.
Access control Five Entra-mapped roles from Consultant down to Visitor, default-restrictive: an unrecognised user sees a summary dashboard only, nothing else.
Unmask groups Dynamic Data Masking on the SQL endpoint: clear values are visible only to an explicit per-client unmask security group, independent of what's physically in silver.
Audit trail Every save, every masking change, every alert send and every schema drift is written to an append-only audit table, not just the pipeline run history.
Data quality Schema drift and failed type casts are captured and reviewable, not silently dropped: a run that succeeds with caveats is flagged as a warning, not a quiet pass.

None of this is a mockup.

The configuration interface, the health dashboard, the run history: all of it is running today. Open it in your browser and click around, no login, nothing to install.

►  Open the live demo

It's already built. The only question is whether it fits you.

Tell us what you're running now and we'll answer that honestly. If it isn't a fit, we'll say so on the call rather than three weeks into a discovery phase.

Book a 30-minute call ►  See it running first

A client partner first.