# Composable Audience Studio Overview

## Introduction to Composable Audience Studio

In the modern enterprise data landscape, cloud data warehouses (CDWs)—primarily Snowflake, Databricks, and BigQuery—have become the central hub for customer information. Composable Audience Studio (CAS) introduces a "Warehouse-First" deployment mode that allows organizations to maximize these investments without the architectural overhead of traditional data replication. Historically, enterprises have struggled to balance data governance with marketing agility: data engineers prioritize a clean "single source of truth" and are concerned about the security risks of data duplication, while marketers face campaign delays of two weeks or more waiting for technical teams to execute SQL queries.

Composable Audience Studio (CAS) resolves this friction by providing a "Zero-Copy" intelligence layer that operates directly on your existing infrastructure.

The strategic value of CAS centers on four key benefits:

- **Activating Data Where It Lives:** Eliminates the cycle of data copying, reduces egress costs, and improves security by keeping batch datasets synchronized within the CDW.
- **Accelerating Time-to-Market:** Empowers marketing teams with self-service, visual segmentation tools to launch campaigns in hours rather than weeks.
- **Unifying Customer Views:** Combines best-in-class data warehousing with Treasure Data’s sophisticated orchestration and CDP intelligence.
- **Future-Proofing the Tech Stack:** Leverages a platform-agnostic framework to ensure long-term architectural flexibility.


This "Zero-Copy" approach transforms the CDW from a static repository into an active engine for customer engagement through a federated compute model.

## Architectural Paradigm: Zero-Copy & Federated Compute

The transition to a Composable architecture represents a fundamental shift away from the traditional model, where data must be fully ingested into a proprietary environment. CAS functions as an **Intelligence & Activation Layer** that operates natively on top of the CDW.

The system utilizes a **Federated Query architecture**. When a marketer builds an audience in the UI, Treasure Data sends a federated query to the customer’s warehouse—leveraging native compute environments such as Snowflake or Databricks—and receives only the results at query time. Data residency is strictly maintained; batch data never leaves the customer’s secure environment until activation for a downstream channel.

The following table differentiates the operational realities of the "Complete" vs. "Composable" modes:

| Feature | Complete Mode (Packaged) | Composable Mode (Warehouse-First) |
|  --- | --- | --- |
| **Data Location** | Ingested into Treasure Data | Resides in CDW (Snowflake, Databricks, BigQuery) |
| **Compute Source** | Treasure Data native compute | CDW native compute (Federated) |
| **Cost Control** | Predictable, flat-fee structure | Variable; organizations must manage CDW usage based on business requirements |
| **Technical Resources** | Low maintenance; TD manages schema | Higher; customer team must design and maintain schemas and tune performance |


This architecture requires a prerequisite data structure within the warehouse, defined through a specific schema and configuration.

In summary, Composable Audience Studio provides a unified path for the enterprise: offering the governance and "Zero-Copy" security required by data engineers alongside the self-service agility demanded by modern marketers.