Configure a data collection environment for web and mobile event capture

The foundational task for any CDP implementation is establishing a reliable pipeline to capture behavioral events from all customer touchpoints. Without this pipeline, profiles cannot be built, audiences cannot be segmented, and activations cannot be personalized.

What this task produces. A published event collection pipeline that captures web page views, product interactions, and authentication events from both a web channel and a mobile channel, routes them through an edge-layer configuration, and delivers them to the customer data store in the correct schema format.

Key configuration decisions. The routing configuration (Datastream in AEP; Kafka topic + consumer group in composable stacks) is the critical decision point: it determines which datasets receive which event types and how governance policy is applied at the edge. Mis-routing events to the wrong dataset is the most common cause of missing behavioral data in early implementations.

Schema alignment. Events must conform to an agreed schema before they reach the data store. In AEP, this is XDM (Experience Data Model). In composable stacks, it is whatever schema the data warehouse table expects — typically defined via dbt source definitions. Getting schema right before going live prevents expensive backfill work.

Parallel viability (medium). Tag management via a TMS (Adobe Tags, Google Tag Manager, Tealium iQ) is broadly applicable. The Datastream construct is AEP-specific; its functional analog in composable stacks is a streaming router (Kafka, Kinesis, Pub/Sub) with a consumer that writes to the CDW. Phase 3 of this task will document the Snowflake/dbt parallel path.