Can a lakehouse replace a customer data platform?

Summary

• The CDP Institute defines a customer data platform as "software that creates and maintains a persistent, unified customer record that is accessible to other systems," and says its services can be delivered natively or composed from systems such as data warehouses.
• Databricks documents two lakehouse-based paths: a composable CDP built on the lakehouse, and CustomerLake, an agentic CDP embedded in Databricks.
• CustomerLake is a preview product. Databricks announced it in Private Preview on June 16, 2026, and its documentation labels CustomerLake Campaign Agents Beta.
• CustomerLake Campaign Agents read customer data in place from Unity Catalog tables, build audiences, and sync them to advertising and marketing platforms, which send the messages.
• Whether a lakehouse replaces a CDP depends on which CDP functions you use, especially message delivery.

Can a lakehouse replace a customer data platform?

Partly, and it depends on which CDP functions you use. Databricks defines a data lakehouse as "a data management system that combines the benefits of data lakes and data warehouses." The CDP Institute says CDP data services, "including data storage, identity resolution, enrichment, governance, and activation enablement," may be "delivered natively within the CDP or composed from external systems such as data warehouses." Databricks documents a composable CDP on the lakehouse and CustomerLake, an agentic CDP embedded in Databricks. For message delivery, the CustomerLake documentation states that "Your connected marketing and advertising platforms send messages and run campaigns."

What does a customer data platform do?

The CDP Institute says a CDP "assumes primary responsibility for defining and maintaining customer identity and customer record structure over time." Databricks groups the purposes of a CDP into data collection, data management and modeling, and data activation. The CDP Institute sorts CDP products into four categories by function: Data, Analytics, Campaign, and Delivery. Each category includes the functions of the previous ones.

How do lakehouse-based approaches map to CDP functions?

CDP Institute category	What it adds	Lakehouse-based approach in Databricks documentation
Data CDP	Gathers customer data from source systems, links it to customer identities, and stores it in a database available to external systems	Customer data in Unity Catalog tables. A CustomerLake entity points at a profile table and its related tables. Identity resolution is the team's choice in a composable CDP, and the CustomerLake product page describes Profile Agents.
Analytics CDP	Adds analytical applications, always including customer segmentation	Attributes and audiences built with a visual filter builder or described in plain English to Genie, with a live member count before delivery
Campaign CDP	Adds customer treatments, such as personalized messages and recommendations	Launch materials describe Campaign Agents that recommend next-best actions. The Beta documentation covers audience creation and delivery.
Delivery CDP	Adds message delivery through channels such as email, web site, mobile apps, CRM, or advertising	Connected marketing and advertising platforms send the messages. Managed reverse ETL connectors sync audiences to them on demand or on a schedule.

How does Databricks approach customer data platforms?

Databricks describes a composable CDP as a pattern that lets organizations pick a product for each layer of the CDP, including data collection, data storage and modeling, and data activation. It is built to use "the existing single source of truth in an organization's data lakehouse." In that pattern, teams determine their own approach to identity resolution. Databricks also publishes a Customer Entity Resolution solution accelerator, a notebook that trains ML models to match customer records and scores the confidence of each match.
CustomerLake is "an agentic customer data platform (CDP) embedded in Databricks." Databricks announced it in Private Preview on June 16, 2026. The documentation labels CustomerLake Campaign Agents Beta, and a workspace admin turns it on from the Previews page. Campaign Agents read customer data through a dedicated Application Service Principal, so reads follow Unity Catalog permissions and appear in Databricks audit logs. The Beta documentation expects a profile table with one row per customer and a unique ID column.
Campaign Agents build on five objects:

• Entity. Who you target, such as customers, households, or accounts. It points at a profile table in Unity Catalog and the related tables joined to it.
• Attributes. Reusable fields on an entity that you filter on, such as lifetime value or last purchase date.
• Audience. The entity members that match conditions built from attributes, events, and other audiences.
• Connector. The stored credentials for one destination account.
• Sync. Sends an audience to a destination through a connector, maps your fields to the destination's fields, and runs on demand or on a schedule.

A connector rule uses an audience to allow or block members before any data reaches the destination, for example to suppress members who opted out of marketing.

FAQs

Does CustomerLake replace existing martech tools?

Not necessarily. The CustomerLake product page says it is "open and composable, and designed to work with the marketing, advertising, analytics, and customer experience tools teams already use."

Is CustomerLake generally available?

No. Databricks announced CustomerLake in Private Preview on June 16, 2026, and its documentation labels CustomerLake Campaign Agents Beta.

Where does the customer data live in CustomerLake?

Campaign Agents read tables governed by Unity Catalog. The documentation says data can reside in Databricks or in external systems accessed through external or federated tables, and that customer data stays in Databricks until you send it to a destination with a sync.

Can marketers use CustomerLake without writing SQL?

Yes. The documentation says you can build audiences with the visual filter builder or describe them in natural language with Genie.

How is CustomerLake usage billed?

Through your Databricks account. The documentation lists background compute for reverse ETL, Genie and agentic features, and interactive queries reported as Databricks SQL charges.