Semantic Layer

Define what 'Revenue' means once, and every dashboard that references it stays consistent — instead of each panel's AI-generated query re-deriving its own definition.

Metrics

A metric is a named, reusable aggregation tied to a datasource and a table — for example:

Metric

Name:        Total Revenue
Table:       orders
Expression:  SUM(o.total_amount)
Description: Sum of all completed order totals in KES

Once confirmed, a metric is injected into the query-generation prompt for that datasource, so any panel whose intent matches "revenue" builds on the same expression instead of the AI inventing its own each time.

Dimensions

A dimension classifies a column by its business role — a datetime, a category, an identifier — along with the grain options it can be grouped by (day, week, month, quarter, year for datetime dimensions). This keeps grouping and filtering consistent across every panel that references the column.

AI-suggested definitions

Run Suggest on a datasource to have InsightBase propose metrics and dimensions from its schema automatically. Suggestions are marked unconfirmed until you review and confirm them — only confirmed metrics and dimensions are used during query generation.

Note:You can also define metrics and dimensions manually instead of relying on suggestions — useful when a calculation is specific to your business logic (e.g. excluding refunds, or a non-obvious churn definition).

Where it's used

•Query generation for new dashboard panels searches confirmed metrics/dimensions before falling back to raw schema.
•Each metric and dimension is tied to one datasource — definitions don’t leak across unrelated datasources.

Reports Team & Workspaces