What Is a Semantic Layer?

A semantic layer is an abstraction between a database and the people who query it, translating business terminology like "revenue" or "active users" into the specific SQL definitions that fetch the right data from the right tables.

Without a semantic layer, different teams often calculate the same metric differently — one using gross revenue, another net. The semantic layer establishes a single source of truth for every metric, dimension, and calculation that the organization uses.

Why the Semantic Layer Matters

Raw databases are built for storage efficiency, not human understanding. Table names like fct_orders_v3 and column names like amt_net_adj are opaque to business users and even to analysts unfamiliar with a particular schema.

The semantic layer solves this by providing a business-friendly vocabulary on top of the raw data. When a user asks for "revenue," the semantic layer knows which table, which column, which filters, and which currency conversion to apply — and it applies them consistently every time.

This consistency is critical as organizations adopt self-service BI. Without it, two people asking the same question can get different answers, eroding trust in data across the company.

How a Semantic Layer Works

1. Metric definitions — Data teams define metrics (e.g., "Monthly Recurring Revenue = SUM of active subscription amounts") with their exact SQL logic, filters, and granularity rules.

2. Dimension mapping — Business-friendly names are mapped to database columns. "Customer region" maps to dim_customer.geo_region, and joins are handled automatically.

3. Access controls — The semantic layer enforces who can see what, applying row-level security and column-level permissions consistently.

4. Query translation — When a user or tool requests a metric, the semantic layer generates the correct SQL, applying the right joins, filters, and aggregations without the user needing to know the underlying schema.

Examples of a Semantic Layer

• Finance: "Gross margin" always means (revenue - COGS) / revenue, regardless of which tool queries it — the semantic layer enforces the definition.
• Product: "Active users" is defined as users who performed at least one core action in the last 30 days — the same definition applies in dashboards, reports, and ad hoc queries.
• Sales: "Qualified pipeline" applies the same stage filters and probability thresholds whether a sales manager asks in a dashboard or through a conversational interface.

Semantic Layer and Lookato

Lookato uses a semantic layer to power its natural language interface. When you ask a question, Lookato maps your business terms to governed metric definitions and database columns — ensuring that "revenue" always means the same thing, no matter who asks or how they phrase it.