How can marketers build audiences without waiting for data engineering?

Summary

• The wait is the ticket queue: audiences usually sit until a data engineer writes SQL and pulls the data. CustomerLake removes that hand-off by letting marketers build and activate audiences themselves in plain language. It is currently in Private Preview.
• CustomerLake is an agentic customer data platform built into Databricks, so marketers work on trusted customer data where it already lives, with no data pull to request and no export into a separate tool.
• Campaign Agents, powered by Genie, turn a plain-English description of a target into a governed audience, then push it to the marketing and activation channels the team already uses.
• The collaboration model changes: data teams define trusted datasets and models once in Unity Catalog, and marketers self-serve on top, so the ad hoc request queue shrinks.
• Every audience is saved as a governed table in Unity Catalog with access control, lineage, and audit, so self-service stays inside the guardrails the data team set.

How can marketers build audiences without waiting for data engineering?

Most audiences wait in a queue. A marketer describes who they want to reach, a data engineer translates it into SQL, pulls the data, and hands back a list, often after the moment has passed. CustomerLake removes that hand-off. It lets marketers build and activate audiences themselves, in plain language, on trusted customer data that already lives in Databricks, so there's no ticket to file and no export to wait on. CustomerLake is currently in Private Preview, so access and capabilities are still rolling out.

What is CustomerLake?

CustomerLake is an agentic customer data platform (CDP) built natively into Databricks. It brings the core CDP jobs (Customer 360 profiles, identity resolution, audience building, and activation) into the lakehouse, where customer data, AI models, and governance already sit. Rather than copying data into a separate marketing system, CustomerLake works on it in place under Unity Catalog. Two agent families do the work: Profile Agents build governed Customer 360 profiles and resolve identity, and Campaign Agents let marketers build audiences and run activation through natural language.

Where the data-engineering wait comes from, and how to remove it

The delay isn't the marketer's fault or the engineer's. It's structural, so removing it takes a few things:

Source of the wait	What removes it
Every audience needs custom SQL	A plain-language interface a marketer can drive without SQL or an engineer.
Data sits in a separate CDP	A CDP built on the lakehouse, so audiences run on data in place with no export.
Trust and access reviews per request	Trusted datasets and models defined once in Unity Catalog and inherited by every audience.
Activation needs another hand-off	Built-in activation that pushes audiences to existing channels with no manual pull.
No safe self-service path	Governed self-service, so access, lineage, and audit hold even when a marketer builds the audience.

How Databricks approaches this

CustomerLake gives marketers a workspace where the entry point is a description, not a query. A marketer states a campaign objective and target profile in plain English, and Campaign Agents, powered by Genie, assemble the audience. Those agents draw on customer attributes, behavioral signals, predictive models, and eligibility rules, so the audience reflects real context, not a rough filter. Marketers can inspect what the agent built, including the generated SQL, affinity summaries, and campaign recommendations, and dig deeper with AI/BI Dashboards embedded in the app.
The collaboration model is the real change. Data teams define trusted datasets and models once through Databricks and Unity Catalog. Marketers then work on top of that foundation through purpose-built interfaces, so the ad hoc request queue that used to sit with the data team shrinks. Built audiences are saved as persisted tables in Unity Catalog, which keeps them governed and auditable, then pushed to the marketing and activation tools and ad networks the team already uses, without exporting data into unmanaged copies. For the product overview, see Introducing CustomerLake, an agentic CDP and the agentic marketing data layer.

Use cases

Self-service audience building. A marketer describes a segment in plain language and gets a governed audience back, with no SQL and no data-team ticket.
Faster campaign iteration. Because building and activating happen in one workspace, marketers can adjust a target and relaunch without opening a new request cycle.
Governed activation. Audiences push to existing channels while staying under Unity Catalog access control, so speed doesn't cost governance.

FAQs

Do marketers need a data engineer to build an audience in CustomerLake?

No. Campaign Agents, powered by Genie, build audiences from a plain-language description, so a marketer can go from idea to activated audience without filing a data-engineering ticket.

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

Yes. CustomerLake runs on the lakehouse, so every audience inherits Unity Catalog access control, lineage, and audit. Self-service does not mean bypassing governance.

Does building an audience require copying data into a separate CDP?

No. CustomerLake is built into Databricks and works on customer data in place, so there is no export into a separate marketing platform to keep in sync.

Can marketers activate the audiences they build?

Yes. Audiences are saved as governed tables and can be pushed to the marketing and activation tools and ad networks the team already uses, without a manual data pull.

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.