All postsThe PerSight Team · Last updated: July 25, 2026 · 5 min read

What Is a Natural Language Query (NLQ)? Reaching Data in Your Own Words

NLQ, or natural language query, is the experience of reaching data with your own sentence instead of SQL or filter panels. How it relates to text-to-SQL, its place in self-service BI, and its limits.

Business intelligence has chased one dream for years: asking the data directly instead of asking the person who knows the report tool. The industry name for that experience is NLQ, short for natural language query.

What exactly is an NLQ?

NLQ is the interface experience where a user can question data with their own sentence: a single question box instead of filter panels, drag-and-drop fields, or a SQL editor. You type "which city's revenue dropped this quarter?" and the answer comes back as a table or chart.

The difference from text-to-SQL is a difference of layers: NLQ names the experience users see, text-to-SQL names the translation engine that makes it possible. Power BI's Q&A feature and PerSight's question box are both NLQ interfaces; underneath each runs some form of natural-language-to-query translation.

Why does it matter now?

NLQ is not a new idea; BI tools shipped keyword-matching versions for years, and the results disappointed. Large language models changed the equation: a layered sentence like "dealers whose revenue last month was lower than the month before" can now be translated correctly. The long-broken promise of self-service BI became realistic for the first time.

How should you evaluate an NLQ experience?

  • Verifiability: does it show the query it generated, or is it a black box? An NLQ whose answers you cannot verify is a fancy guessing machine.
  • Data safety: where do your question and your data go? Is the schema sent, or actual rows?
  • Read-only guarantee: can natural language trigger a write or delete? If it can, walk away.
  • Language support: does it truly work in your language, or only in English?

Use it with its limits in mind, too: NLQ has to guess on ambiguous questions ("best products" by what measure?), and complex analytical flows (cohort analysis, multi-step funnels) still need an analyst's design. NLQ is the tool for everyday questions stuck in the report queue, not a replacement for the data team.

Frequently asked questions

Are NLQ and text-to-SQL the same thing?
Intertwined but not the same: NLQ names the user experience, text-to-SQL the translation technique powering it. Most NLQ products run text-to-SQL as the engine.
Do I need a data warehouse or BI setup for NLQ?
No. Some tools sit on top of an existing BI platform; tools like PerSight connect directly to your database. You can start with the tables you have today.
Does NLQ work in languages other than English?
With modern models, yes. A Turkish or German question matches even English column names. PerSight treats Turkish as a first-class language alongside English and German.
What is NLQ not suited for?
Fixed-template scheduled reports, multi-step analytical constructions, and data modeling are not NLQ's job. NLQ is for instant question-and-answer: "how many orders yesterday", "which product stood out this month".

If you want to test an NLQ against the four evaluation questions above, PerSight's setup is: a question box in your language, generated SQL always visible, read-only execution, and row data that never leaves your machine. persight.ai/beta

References