Google Cloud has invested heavily in Apache Iceberg, and recently launched its cross-cloud Lakehouse, which brings BigQuery’s performance and governance to open lakehouse data. Zeotap’s Composable CDP now runs on that foundation. Marketing, growth, and advertising teams can build audiences, resolve identities, and activate customer data directly on their Iceberg lakehouse, on the same tables their analytics and data science teams already use. The platform is also agentic, so those teams can plan and run all of it through an on-platform agent or from their own AI tools.
For most enterprises, customer data has lived in two places: a governed copy in the warehouse or lakehouse, and a second copy inside the CDP where marketing could act on it. Keeping those two in sync has always been work, and the definitions rarely stayed identical. Running on an open lakehouse removes the second copy. Marketing works on the same governed data as analytics and AI, under the same controls.
Why an open lakehouse matters for marketing
As enterprises modernise onto lakehouse architectures, most are standardising on Apache Iceberg and other open table formats. Two properties make that shift useful for marketing teams specifically.
- One open copy, many engines. The same Iceberg table can be read and written by BigQuery, Spark, Trino, and other engines. There is no extra copy to maintain and nothing trapped in a proprietary format, so the data your analytics team curates is the data your marketing team activates.
- Composable by design. Storage is separate from compute, so you can choose the right engine, or the right application, for each workload and change it later. Your activation layer can evolve without re-platforming the data underneath it.
For marketing, that means the work of defining a customer, scoring an audience, or adding an attribute happens once, on data everyone shares, rather than twice across two systems that drift apart.
How Zeotap’s Composable CDP runs on the Lakehouse
Zeotap works on your Iceberg tables in place. It queries them through BigQuery and writes back to the same tables, with no replication into a separate customer data store. Your Google Cloud governance, access controls, and lineage stay intact, because the data never leaves your environment. And because BigQuery can read Iceberg data across clouds through Google Cloud’s cross-cloud Lakehouse, Zeotap can work with that data wherever it sits, including in AWS or Azure.
This works with the Google Cloud setup you already have. There is no new data store to secure or pay for, and no migration project before you can start. If your data is already in BigQuery, or in Iceberg tables governed by Google Cloud’s Lakehouse, Zeotap connects to it directly and sits on top of your existing data foundation rather than beside it.
What you can run on the lakehouse
On that foundation, your teams have the full Composable CDP, working directly on your Iceberg tables.
Unified profiles and identity. Collect first-party event data into the lakehouse and resolve it into unified customer profiles, using both deterministic and probabilistic identity resolution. The resolved profiles live on your governed tables, not in a separate identity store, so the customer your analysts see is the customer your marketers target.
Audiences, segments, and journeys. Build audiences, segments, and multi-step journeys against the same governed tables. When the data team publishes a new attribute or model score, it is available to marketing immediately, with no pipeline to copy it across first. The cycle from a new idea to a live audience shortens from weeks to the time it takes to write the segment.
Activation and match rates. Activate to 250+ destinations across paid media, ad platforms, CRM, marketing automation, and on-site and in-app personalisation. Audience Boost raises paid-media match rates, applied on top of your first-party data without moving it out of your environment, so more of every audience reaches its intended platform.
Consent and governance. Consent is applied at the point of activation, so every audience honours the permissions attached to the underlying data. Because activation runs on governed tables in your own cloud, there is no shadow copy to audit and no second definition of a customer to reconcile. Your existing data controls cover marketing the same way they cover analytics.
An Agentic Composable CDP
Zeotap’s Composable CDP is also agentic. An on-platform agent lets your teams plan, iterate on, and operate every capability above, from resolving identities and building audiences to running journeys and activating to your destinations. It takes on much of the planning and operational work that normally sits between an idea and a launched campaign, which removes a real source of friction and lets your teams work faster and get more done.
The same capabilities are open off-platform too. Zeotap exposes its full CDP capabilities through the Model Context Protocol (MCP), so your teams can drive identity, audiences, journeys, and activation from the AI tools they already use, including Claude and Gemini. An enterprise-grade AI governance layer sits in front of that access, so your permissions and consent rules apply whether a capability runs inside Zeotap or from an outside tool.
Get started
If your customer data already lives on Google Cloud’s Lakehouse, or as Apache Iceberg tables in BigQuery, Zeotap’s Composable CDP can activate it now. To see it against your own data, book a 30-minute working session with our solutions team.
