Natural language query

Natural language query lets someone type or speak a question in ordinary language—"what were sales in the Northeast last quarter?"—and receive an answer, typically a number, chart, or short summary, without writing SQL or navigating a BI tool's menus. It is the interface layer of a broader move toward augmented analytics, where AI handles the translation between a business question and the underlying data.

Under the hood, most natural language query systems rely on a language model to parse intent and map it onto a defined data structure—tables, metrics, and dimensions the system already knows about—then execute a query and return the result in plain language. This underlying translation step is closely related to Text-to-SQL, though NLQ systems often add guardrails, such as a governed semantic layer, to keep answers consistent with an organization's official metric definitions.

Natural language query is used to extend Self-service BI to people who do not know SQL or a BI tool's query builder, lowering the barrier to ad hoc questions. Its main pitfalls are answering confidently with the wrong interpretation of an ambiguous question, and returning a plausible number that does not match the organization's official metric definition because the system used its own logic instead of a governed one—both argue for showing the underlying query alongside the answer, not just the result.