How can marketers explore customer data and create segments without knowing SQL?

Summary

• Marketers explore customer data by asking questions in plain English through Genie, then turn what they find into segments, with no SQL required. In CustomerLake this runs on governed data in Databricks, which is currently in Private Preview.
• Genie is a data agent that translates a plain-language question into SQL, runs it, and returns both the query and a results table, so a marketer can probe behavior, channel preference, or churn risk across the full depth of customer data, not a pre-aggregated slice.
• Exploration and segmentation are one flow: the patterns a marketer surfaces by asking questions become the criteria for an audience, described in words rather than written as a query.
• CustomerLake adds Genie-powered natural-language segmentation on top of a governed Customer 360, so segments run where the data already lives, with no copy into a separate CDP.
• Because it runs on the lakehouse, every question and segment inherits Unity Catalog permissions and lineage, so a marketer only explores data they're cleared to see.

How can marketers explore customer data and create segments without knowing SQL?

A marketer doesn't need to write SQL to understand customers or build a segment. With Genie, Databricks' data agent, you ask a question in plain English, and Genie translates it into a query, runs it, and shows you the result. That turns exploration into a conversation: ask what drives churn in a region, read the answer, refine the question, and the moment you know the group you want, describe it and save it as a segment. In CustomerLake, this runs on governed customer data inside Databricks. CustomerLake is currently in Private Preview, so access and capabilities are still rolling out.

What is natural-language data exploration?

Natural-language exploration lets a business user ask a question in ordinary words and get a data-backed answer, with an AI agent handling the query. Genie is configured by the people who know the data: domain experts point it at the right datasets and give it sample queries and written guidance. A marketer then asks in plain English, Genie generates the SQL, runs it, and returns both the query and a results table. If a question is ambiguous, Genie can ask a follow-up to clarify. Because the generated SQL is shown, the marketer can reuse it and an analyst can check it.

From exploring to segmenting: what to look for

Doing both without SQL takes a few capabilities working together:

Capability	Why it matters
Plain-language questions over full-depth data	A marketer should query raw customer behavior, not a pre-built dashboard slice, so answers reflect the real population.
Visible, reusable query logic	Showing the generated SQL and a result preview lets a marketer trust the answer and an analyst verify it.
One path from insight to segment	The criteria a marketer discovers by asking questions should convert directly into an audience, with no rebuild.
Governed access	Exploration and segments must respect the permissions and lineage the data team set, so a user only sees data they're cleared for.
Segments on data in place	Running segments where the data lives avoids copies that drift and escape governance.

How Databricks approaches this

Genie is the exploration surface. A marketer asks about segment behavior, loyalty performance, channel preference, or churn risk in plain English, and Genie answers from governed data, drawing on the full depth of customer records rather than a pre-aggregated report. For how Genie interprets questions and returns generated SQL, see the Genie Agents concepts.
CustomerLake carries that same plain-language model into segmentation. It is an agentic customer data platform built into Databricks, with Genie-powered natural-language segmentation, so a marketer describes an audience in words and gets it back with no SQL. Profile Agents build a unified Customer 360 from first- and third-party data without moving or duplicating it, and Campaign Agents handle segmentation, personalization, and activation. Because segments run on governed data where it already lives, there's no copy into a separate CDP to keep in sync. The result is one collaboration model: data teams define trusted datasets and models once in Unity Catalog, and marketers explore and segment on top of them. See Introducing CustomerLake, an agentic CDP and how Databricks supports customer segmentation.

Use cases

Churn investigation to win-back audience. A marketer asks which customers are lapsing and why, sees the pattern, then saves the at-risk group as a segment for a retention campaign.
Channel and offer analysis. Ask which segments respond on which channels, then build the audience that matches, all in plain language.
Ad hoc questions without a ticket. Rather than requesting a report, a marketer asks Genie directly and gets a governed, data-backed answer.

FAQs

Do marketers need to know SQL to explore data or build segments?

No. Genie turns a plain-English question into SQL, runs it, and returns the answer, and CustomerLake's Genie-powered segmentation builds audiences from a plain-language description. Neither step requires the marketer to write SQL.

Can a marketer check what the agent actually queried?

Yes. Genie shows the SQL it generated along with the results, so a marketer can reuse it and an analyst can verify the logic.

How does exploration turn into a segment?

The criteria a marketer discovers by asking questions become the definition of an audience. In CustomerLake, they describe that audience in words and save it as a governed segment.

Is customer data still governed when a marketer self-serves?

Yes. Questions and segments run on the lakehouse under Unity Catalog, so access control and lineage apply, and a marketer only sees data they're authorized to query.

Is CustomerLake generally available?

No. CustomerLake is currently in Private Preview, so availability and capabilities are still rolling out. Talk to your Databricks account team about access.