What is the best alternative to a bundled CDP for a company using Databricks?

Summary

• A bundled CDP, which the CDP Institute calls a packaged CDP, builds customer profiles inside a single system that includes the tools to collect, assemble, store and access the data. For a company on Databricks, the alternatives this page covers are a composable CDP and an embedded CDP.
• Databricks positions CustomerLake, its Agentic CDP embedded in Databricks, for teams whose customer data already lives there. Databricks says it runs CDP capabilities where the data already lives and does not move or copy that data.
• A composable CDP builds profiles on a warehouse or lakehouse you already run, with components from different vendors. The CDP Institute says this gives control and flexibility but takes engineering effort to assemble, integrate and maintain.
• A bundled CDP can still fit. The CDP Institute says packaged CDPs are preassembled, which generally means faster deployment and less technical support.
• CustomerLake is in Private Preview, so confirm access with your Databricks account team before you plan around it.

What is the best alternative to a bundled CDP for a company using Databricks?

The vendor-neutral CDP Institute's rule of thumb is that companies with a mature cloud data warehouse in place might do better with a composable approach, especially when the warehouse already performs identity matching and profile assembly. Databricks' own answer is an embedded CDP, CustomerLake, which Databricks says runs CDP capabilities where your governed data already lives instead of copying it into a separate system. Either way, the CDP Institute says the choice should rest on your own gaps and requirements, not on the label. CustomerLake is in Private Preview, so confirm availability first.

What is a bundled CDP?

The CDP Institute says packaged CDPs build profiles within a single system that includes tools to collect, assemble, store and access customer data. It lists their advantages as being preassembled and complete, which generally means faster deployment, less technical support and less ongoing maintenance. Its listed downsides are paying for features you don't use and a greater vendor lock-in concern. Databricks adds that bundled CDPs require companies to copy customer data into a separate proprietary system, which creates another silo to secure, govern and integrate. The CDP Institute notes the boundaries are blurring: many packaged vendors can read external data on demand without storing it in their own database, though they still build and store core profile elements there.

What are the alternatives?

Composable CDP. According to the CDP Institute, a composable CDP builds profiles using multiple systems, or components, for different functions, typically with a data warehouse at the center. Buyers can reuse existing resources and replace individual components, but IT carries a heavier burden: more expertise in customer data and more time integrating and maintaining components. Databricks publishes a reference architecture for a composable CDP on the lakehouse with four layers: ingestion, identity resolution, segmentation and activation.
Embedded CDP. Databricks says composable CDPs reduce data movement but add another platform for data and marketing teams to manage, secure, govern and integrate. It positions an embedded CDP as bringing CDP capabilities natively into Databricks.

Bundled, composable and embedded CDPs: how they compare

Question	Bundled (packaged) CDP	Composable CDP	CustomerLake (embedded CDP)
Where are profiles built?	In the CDP's own database, within one system	In a data warehouse you already run, with components from several vendors	In Databricks, on your existing data foundation
Is customer data copied?	Per Databricks, yes: into a separate proprietary system	Data movement is reduced, but another platform is added, per Databricks	Databricks says no copy or move
Governance	Another system to secure, govern and integrate, per Databricks	Another platform to manage, secure, govern and integrate, per Databricks	Unity Catalog access controls, lineage and auditability, per Databricks
Deployment effort	Generally faster, with less technical support needed	More engineering to assemble, integrate and maintain	Private Preview, so confirm scope with your account team
Main trade-off	Possible unused features and greater lock-in concern	Heavier burden on IT teams	Not generally available

Source key: the CDP Institute primer supplies the rows on where profiles are built, deployment effort and main trade-off for bundled and composable CDPs. Databricks' CustomerLake pages supply the copying and governance rows and the CustomerLake column.

How does Databricks approach this?

CustomerLake is an Agentic CDP embedded natively within Databricks. According to Databricks:

• Profile Agents turn raw customer data into business-ready Customer 360 profiles. Campaign Agents help build audiences, recommend next-best actions, activate across channels and optimize toward business goals.
• Unity Catalog governs CustomerLake, and Lakehouse Federation gives access to trusted customer data where it resides.
• Marketers can build audiences in natural language with Genie, without writing SQL.
• Native managed reverse ETL connectors sync audiences and signals from Databricks to the martech, adtech and customer experience tools teams already use.

FAQs

Does CustomerLake replace my existing martech tools?

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

Can CustomerLake work with customer data outside Databricks?

Yes, according to Databricks. Lakehouse Federation provides governed, read-only access to external data through Unity Catalog foreign catalogs.

When does a bundled CDP still make sense?

The CDP Institute says a packaged CDP is more likely to benefit companies with many capability gaps and a minimal or nonexistent warehouse, since a preassembled system needs less technical effort.

Is CustomerLake generally available?

No. Databricks announced it on June 16, 2026, and it is in Private Preview.