The Water Analogy: Creating a Semantic Layer in Microsoft Fabric

How raw business data becomes trustworthy, consistent insight — and why that matters

Why Reports Disagree

If you've ever sat in a meeting where two teams presented different revenue numbers for the same time period — and both were technically correct — you've experienced the core problem that a semantic layer exists to solve.

The issue isn't bad data. It's that different people are measuring the same thing differently, pulling from different places, or applying different logic. The result is mistrust, rework, and decisions made on shaky ground.

A semantic model solves this by establishing a single, governed definition for your business metrics. Understanding what that means and why it matters starts with a simple analogy.

Data Is Like Water

Imagine that all of your business data is water.

Water doesn't have a fixed shape — it takes the shape of whatever container holds it. In data terms, those containers are tables. The structure of a table determines what the data looks like, how it behaves, and how it can be used. This is what data professionals mean when they talk about data shape.

When a company transitions from one system to another, it has an opportunity to return to the cupboard and choose better containers — ones shaped more intentionally for how the business actually needs to use the data. Better-shaped containers mean data that is more consistent, more trustworthy, and more useful for decision-making.

But shape alone isn't enough. Before that water reaches the tap, it needs to travel through a reliable system — one that collects it, cleans it, and delivers it in a form people can trust.

How Data Travels: From Source to Report

Data doesn't go directly from a source system into a dashboard. It moves through three distinct stages, each with a specific responsibility.

 Source Systems (ERP, CRM, Spreadsheets, etc.)
        │
        ▼
 ┌─────────────────────────────────────────────┐
 │  Bronze Layer — Capture                     │
 │  Raw data lands here, unchanged.            │
 │  Every record, exactly as received.         │
 └──────────────────────┬──────────────────────┘
                        │
                        ▼
 ┌─────────────────────────────────────────────┐
 │  Silver Layer — Cleanse                     │
 │  Duplicates removed. Formats standardized.  │
 │  Data from different systems merged.        │
 └──────────────────────┬──────────────────────┘
                        │
                        ▼
 ┌─────────────────────────────────────────────┐
 │  Gold Layer — Deliver                       │
 │  Business definitions locked in.            │
 │  Metrics calculated. Semantic model lives   │
 │  here.                                      │
 └──────────────────────┬──────────────────────┘
                        │
                        ▼
            Reports & Dashboards

Bronze is the raw data layer. Think of it as buckets that catch every drop of water exactly as it falls from the source — no filtering, no interpretation, no reshaping. Its job is preservation. If something ever looks wrong downstream, Bronze is where you trace it back to. Nothing is ever altered here.

Silver is where the cleaning happens. Duplicate records are removed. Inconsistent formats are standardized across systems. Data from your ERP, CRM, and other tools is merged into unified, coherent tables. The water is filtered and combined, but it hasn't been bottled yet.

Gold is where data becomes business-ready. This is where metrics are calculated, definitions are enforced, and the semantic model lives. It is designed entirely around the needs of the people consuming it.

The Semantic Model: Your Single Source of Truth

The semantic layer — built in Microsoft Fabric — is the business logic that sits between your data and your reports.

Think of it as the label on a water bottle. The label tells you exactly what's inside, where it came from, and who it's for. You don't need to understand the treatment process to trust what you're drinking. A semantic model does the same thing for data.

When a semantic model is in place:

"Revenue" means the same thing in every report — the same formula, the same filters, every time.
"Active Customers" has one definition, not five different ones across five different dashboards.
Relationships between tables are defined once and reused everywhere, rather than rebuilt inside each individual report.
Row-level security can be enforced at the model level, so the right people see the right data automatically.

Without a semantic model, every analyst who builds a report has to interpret the data themselves — deciding what counts as revenue, how to join tables, what to exclude. Multiply that across a team, and you get inconsistent numbers, reconciliation headaches, and eroded trust in the data.

With a semantic model, those decisions are made once, reviewed, and locked in. Anyone building a report connects to the same foundation.

What This Looks Like in Practice

Without a Semantic Layer	With a Semantic Layer
Revenue looks different in each report	Revenue is defined once and shared everywhere
Analysts spend time reconciling numbers	Analysts spend time on analysis
Business logic lives inside individual reports	Business logic lives in a governed, central model
New reports require rebuilding definitions	New reports inherit existing, validated definitions
Trusting a number means knowing who built the report	Trust is built into the layer itself

Why This Matters

A semantic layer isn't a technical nicety — it's a prerequisite for trusting your data at scale.

When business metrics are defined centrally and governed consistently:

Executives see the same numbers whether they're in a dashboard, a report, or a one-off data export.
Teams stop arguing about whose figures are right and focus on what the figures mean.
New reports are faster to build because the shared definitions already exist.
Audits and compliance are cleaner because data lineage traces clearly from source system to final report.

The water analogy captures it simply: if Bronze is about collecting water and Silver is about cleaning it, Gold is about bottling it with a clear label and placing it on the shelf — ready for anyone to pick up and trust without needing to understand the plumbing behind it.

That's what a semantic model does for your business. It turns data infrastructure into a reliable utility — one where the answer to "what does this number mean?" is the same, every time, for everyone.