Composable CDP

Data & Tracking

Also: Composable Customer Data Platform · Warehouse-Native CDP

What it isCDP built on your own data warehouse

Versus traditionalNo separate copy of your data

Watch forNeeds real data engineering support

Pairs withReverse ETL and a data warehouse

Quick definition

A Composable Customer Data Platform (CDP) is a customer data setup built directly on top of a business's existing data warehouse rather than a separate proprietary database. Instead of copying customer data into new storage, tools sync data out of the warehouse into marketing and sales platforms using reverse ETL. Fewer copies, fewer sync delays.

How it varies across Australia

Composable CDP adoption in Australia sits mostly with larger businesses that already have a data warehouse and a data engineering function. Mid-market Australian companies more often still run traditional, all-in-one CDPs or no CDP at all, because standing up a composable stack needs technical resources most marketing teams don't have on hand.

See data and tracking maturity across Australian industries →

What it actually means

Picture two translators. One insists on keeping a copy of every document you hand over, filing it in their own cabinet, and speaking for you from that copy forever after. The other works straight from your original documents, wherever they live, and never makes a duplicate. A traditional customer data platform (CDP) is the first translator. A composable CDP is the second.

Instead of pulling customer records out of your data warehouse and into a new proprietary database, a composable CDP treats the warehouse (Snowflake, BigQuery, Databricks) as the single source of truth. Segmentation, identity resolution and audience building happen on data that stays where it already lives. Activation into tools like email platforms, ad networks or a CRM happens through reverse ETL, which pushes warehouse data out to the destination instead of pulling it in.

The appeal is obvious once you've dealt with a traditional CDP's sync lag or paid twice to store the same customer record in two systems. The catch is just as obvious. A composable CDP assumes you already have a well-modelled warehouse and someone capable of maintaining it. Buy the wrong architecture for your team's maturity and you've traded a vendor problem for an engineering problem.

A composable CDP isn't a product you buy. It's an architecture decision that assumes you already have a data warehouse worth building on.

How it shows up

Composable CDPs show up as a set of tools rather than one dashboard. A reverse ETL layer like Hightouch or Census sits between the warehouse and destinations. Identity resolution and segmentation logic often live as SQL models or dbt transformations rather than a vendor's proprietary rules engine. Marketing teams typically request a segment through a form or a Slack request rather than building it visually, because the underlying models live with the data team.

It also shows up in the sales pitch itself. Vendors selling into this space talk about 'activation' and 'the modern data stack' rather than 'unified customer profiles', because the profile was never theirs to unify.

The Australian context

Australian businesses running a composable CDP are almost always ones that adopted a cloud data warehouse early, usually retailers, finance businesses or scale-ups with an existing analytics function. For everyone else, the barrier isn't the concept, it's the headcount. A composable stack needs someone who can write and maintain SQL models, which most Australian marketing teams don't budget for separately from a CDP subscription.

Where people get this wrong

Buying reverse ETL tooling before the warehouse data is properly modelled.Reverse ETL only activates what's already clean in the warehouse. Messy source data produces messy segments faster, not cleaner ones.

Assuming composable means cheaper.The licence cost often drops but the engineering cost rises. Total cost of ownership can land higher once you count the salary of whoever maintains the pipelines.

Choosing composable purely to avoid vendor lock-in.You've swapped lock-in to a CDP vendor for lock-in to your own warehouse schema and whoever built it. That's still a dependency, just an internal one.

Common questions

Do I need a data warehouse before considering a composable CDP?

Yes. A composable CDP has nothing to compose without a warehouse already holding clean, well-modelled customer data. If you don't have one, evaluate a traditional CDP first or build the warehouse before the activation layer.

Is a composable CDP cheaper than a traditional CDP?

Licence fees are usually lower but the engineering cost of building and maintaining the pipelines is real and ongoing. For businesses without an existing data team, total cost often ends up similar or higher than a traditional CDP.

What is reverse ETL and why does it matter here?

Reverse ETL moves data out of a warehouse into operational tools like email platforms or CRMs, the opposite direction of a standard ETL pipeline. It's the piece that lets a composable CDP activate warehouse data without duplicating it into a new system.

Which businesses should avoid composable CDPs?

Any business without an existing data warehouse or a data engineer to maintain it. Composable architecture assumes technical maturity that most small and mid-market marketing teams haven't built yet. A traditional CDP is usually the better starting point.

Debrief

Get the next one

No spam. No fluff. Just the next article, straight to your inbox.

Keep exploring

About New Rebellion

New Rebellion is a marketing intelligence consultancy. We build tools, score Australian businesses on how their marketing actually performs, and publish Debrief every day. This dictionary is part of how we work in the open.

How we think →