Composable vs Packaged CDPs: With Databricks and Braze

For a significant portion of my time at Shaw/Scott, I’ve been working in Databricks on a daily basis, building solutions that activate marketing data within Braze. Recently, at an industry conference, a marketer asked me: “We have Databricks, we have Segment, and we have Braze—but how do we actually make these work together?”

We operate in a marketing landscape where data is central to success. It is well established that highly personalised, conversational marketing delivers stronger results. As a result, having the right technology stack is critical. This article provides some background on CDPs and explores where Databricks and Braze can most effectively fit within a modern martech stack.

Traditional (Packaged) CDPs

Traditional CDPs have supported marketers for many years. They simplify the process of collecting, unifying, and activating customer data, enabling teams to build more advanced solutions quickly. They also support broader marketing needs such as reporting and campaign execution.

A key advantage of traditional CDPs is that they are pre-packaged. This makes them relatively quick to implement and start delivering value. Because the functionality is already built, teams can focus on integrating their data and activating use cases, rather than building infrastructure from scratch.

However, selecting the right CDP is not straightforward. Most platforms began with a specific strength and expanded over time. While they broadly offer similar capabilities, each tends to excel in different areas. Some are particularly strong at event streaming and data movement, others at identity resolution, and others at segmentation and audience building.

Choosing the right platform depends on aligning these strengths with your team’s priorities.

That said, traditional CDPs introduce an additional data layer to manage. In a GDPR-conscious environment, this can create approval hurdles and increase operational overhead.

This raises an important question: where does a traditional CDP sit alongside platforms like Databricks? And with tools like Braze offering increasingly rich functionality, is a traditional CDP always necessary?

Composable CDPs

For many organisations, maintaining an additional data platform is only justifiable if it delivers clear incremental value. In some cases, only a subset of CDP functionality is actually required. This is where composable CDPs become a compelling alternative.

Composable CDPs are built on top of your cloud data warehouse, using a modular approach. Rather than relying on a single platform, you integrate best-in-class components to address specific needs. For example, if identity resolution is your primary challenge, you can integrate a specialised solution directly with your data warehouse using a no-copy architecture.

This approach allows you to keep your data centralised while designing a CDP tailored to your organisation. Although it requires more upfront effort than a packaged solution, it provides greater flexibility and control. It also avoids duplicating data, which reduces compliance risk, lowers operational overhead, and improves governance.

Using Braze

Platforms like Braze enable near real-time integration with cloud data warehouses. This allows you to update attributes, trigger events, and make API calls directly from your data environment—positioning the warehouse as the core of your marketing stack.

A key decision in a composable architecture is defining Braze’s role. Many of its capabilities—such as segmentation and audience management—overlap with traditional CDPs. To avoid duplication and confusion, it is important to establish clear governance: which platform is responsible for which function, and in which contexts.

Where Braze should take clear ownership is in activation and personalisation. Features such as segmentation, Liquid templating, canvases, connected content, and context-driven messaging are designed for orchestrating highly personalised, multi-channel experiences.

By feeding Braze well-structured, pre-aggregated data from your warehouse, marketers can operate more independently without constant reliance on technical teams.

With strong integration and thoughtful data modelling, many marketing use cases can be delivered directly through this setup. While there will still be occasions where additional data engineering support is required, good upfront design can minimise these dependencies.

Where Databricks Fits

Databricks serves as the foundation of a composable CDP, acting as the central data platform. It can also be integrated with traditional CDPs, either to provide a more user-friendly interface for marketers or to leverage specific CDP capabilities.

From experience, Databricks is particularly well suited to this role for several reasons:

• It offers extensive flexibility in reading and writing multiple data formats, along with strong connectivity to storage and compute systems.

• Its marketplace and partner ecosystem allow you to quickly integrate validated tools, including CDPs.

• Delta Sharing enables zero-copy data collaboration, allowing integrations without duplicating or moving data.

• Unity Catalog provides secure data governance across teams while maintaining appropriate isolation.

  
  

In some cases, integrating Databricks with a traditional CDP can still be beneficial. This is particularly relevant if your marketing tools lack sufficient aggregation or segmentation capabilities, but you want to empower more data-literate marketers without requiring them to work directly in the data warehouse.

Conclusion

Returning to the original question—how do Databricks, Segment, and Braze work together?—the reality is that there is no single answer.

Every organisation differs across several dimensions:

• Industry

• Strategic and tactical marketing goals

• Technical capability of the team

• Relationship with the wider business

• Internal processes

• Budget

• Available time

  
  

When I spoke with this marketer, the conversation focused less on tools and more on their organisation, use cases, and constraints. As a general guide:

• Highly technical marketing teams may be able to operate directly from Databricks, provided strong governance and deployment processes are in place.

• Working with an agency such as Shaw/Scott can help bridge the gap, managing both the data platform and activation layer while designing effective pipelines.

• If you require specialised capabilities such as identity resolution, integrating a traditional CDP into a composable architecture via marketplace connectors can be effective.

• If empowering marketers with more hands-on control is a priority, surfacing curated data into a CDP layer may be the right approach.

  
  

There are no universal answers—but there are well-informed approaches.

At Shaw/Scott, we have extensive experience across data platforms, CDPs, and marketing technologies, and how they integrate in practice. We typically recommend starting with an audit of your team, use cases, and requirements to define the optimal architecture.

If you’re evaluating your martech stack and aren’t sure where to begin, we’d be happy to help.