Customer data platforms have long been the backbone of effective digital advertising - but not all CDPs are built the same. A newer approach - the composable CDP - lets brands keep their data in their own warehouse and build on top of it, rather than handing it to a vendor. Niko Batinica, a member of the IAB New Zealand Emerging Technology Council and Partner at Team Circle explains composable CDPs and why they matter.
Composable CDP: What is it and why should I care?
Your customer data dictates everything. It determines your advertising efficiency, your ad platform match rates, your ability to deploy useful AI, and whether your campaigns stay profitable in a stricter, privacy-first world. But to turn that scattered data into a competitive advantage, you must get your underlying architecture right. That brings us to the foundation of modern data strategy.
What is a CDP at all?
A Customer Data Platform (CDP) is a collection of your customer data from different touchpoints (think phone calls, website visits, email clicks, in-store purchases). This data is ideally tied into a single, unified profile with an interface sitting over the top that helps make that data usable. Getting this right means the difference between treating your most loyal buyers like VIPs or treating them like complete strangers every time they click an ad.
A CDP is made up of three broad parts:
Historically, companies bought a “Packaged” CDP. This is your traditional, all-in-one service. You send your data to the vendor, and they provide a user-friendly interface to build audiences and launch campaigns.
This style of traditional CDP can still be a fantastic choice. It streamlines setup, provides out-of-the-box simplicity, and tends to be more marketing friendly. However, as organizations scale, the closed nature of these platforms can lead to siloed data and steep pricing curves. You are also reliant on the vendor roadmap for features and integrations.
Okay, then what is a Composable CDP?
As companies grow, they tend to build their own central database, known as a cloud data warehouse (CDW). This inevitably involves a large amount of customer data.
A Composable CDP simply means you leave your customer data in your own warehouse. Instead of moving it to a new platform, you build logic and develop tools directly on top of your CDW. This can be anything from unification logic for users, tools to organise the data, or connectors to activate it in advertising platforms.
You don't copy the data into a vendor's black box. You assemble (or compose) the pieces you need. This introduces a vital concept called 'zero-copy architecture'. Instead of duplicating your data (exposing yourself to security risks and paying to store it twice) your marketing tools (for example Hightouch or Census) simply read the data directly from where it already lives. Many traditional CDPs (like Salesforce) are also starting to integrate this architecture but again you are reliant on them having connectors available.
Why does this matter for advertisers?
Everyone is feeling the pinch of signal loss and the death of third-party cookies. First-party data (the data you actually own) is the most valuable currency left in marketing. To maintain high match rates on ad platforms, you need a strong data foundation.
Everyone is feeling the pinch of signal loss and the death of third-party cookies. First-party data (the data you actually own) is the most valuable currency left in marketing. To maintain high match rates on ad platforms, you need a strong data foundation" - Niko Batinica, Partner Team Circle,
Say you are a retailer running retargeting campaigns. With a traditional setup, a lag in data syncing means you’ve spent ad dollars serving a cart-abandonment ad to a customer who actually bought the item in-store an hour ago. However reading directly from the CDW, in a composable set up, you can instantly suppress that recent purchaser, saving ad spend and improving the customer experience.
There is also a massive cost efficiency play here. Packaged CDPs often charge based on the total number of customer profiles or activity usage that accounts for the cloud computing power the software uses, plus margin. Composable architecture allows you to store data cheaply in your own warehouse and only pay for the computing power you actually use.
The AI Evolution
In the past, composing your own data stack meant hiring an army of expensive data engineers. Generative AI has changed the math entirely.
AI coding assistants have systematically collapsed the cost and friction of custom development. Because building is now faster and cheaper, the old 'build vs. buy' debate is dead. As the 2026 State of Martech report notes, smart organizations do both. They buy standard tools for the basics and use AI to build custom components for their unique needs.
More importantly, this changes how you use marketing AI. Because your data stays in your own warehouse, you can plug in predictive AI models to score leads, build lookalike audiences based on lifetime value, or predict churn. You no longer have to wait for a CDP vendor to build those specific features into their roadmap.
First-Party Data and the Agentic Future
The massive growth of Retail Media Networks (a $50.5 billion sector globally*) proves that owned customer data is the definitive future of digital advertising.
Looking ahead, your CDP will need to do more than serve humans. It will serve autonomous AI agents. The real challenge of tomorrow isn't generating content, it's 'context engineering'. AI is only as smart as the data it has access to. Context engineering means providing AI with precise, governed data and rules so it doesn't accidentally offer a 20% discount to a VIP customer who just happily paid full price.
Which path is right for you?
The balance here is relying on an out-of-the-box solution from a traditional CDP or building a custom solution for yourself.
Choose a Packaged CDP if: You want a less technical solution, need a turnkey solution launched rapidly, and have relatively straightforward marketing channels. If you do, make sure you understand what you are paying for, how to manage cost and what in the platform is delivering value.
Choose a Composable CDP if: You already heavily utilise a cloud data warehouse, require custom predictive modelling, and want to eliminate data duplication costs. If you do, be clear around requirements, focus on demonstrating tangible value and plan your architecture to avoid rework
The Bottom Line
You already own your data, and you likely own the cloud warehouse where it lives. With AI completely collapsing the cost of custom development and zero-copy architecture minimising the security risks of moving data, the power has shifted back to the brand.
Stop asking whether you should strictly build or buy. Start orchestrating a flexible data ecosystem that protects customer privacy, lowers your operational costs, and turns your existing infrastructure into an engine for growth.
*https://www.grandviewresearch.com/industry-analysis/retail-media-networks-market-report
This article was written by Niko Batinica, Partner at Team Circle and member of the IAB New Zealand Emerging Technology Council.
For further information or to become an IAB New Zealand Member, please get in touch.