Data Management and Segmentation
Core mental model
Adobe Experience Platform data model: schemas (XDM), datasets, identities and identity namespaces, merge policies, and the Real-Time Customer Profile. Segmentation evaluates segment definitions against the profile store; streaming segmentation updates membership near real-time, batch segmentation runs on a schedule.
Concepts candidates must know
- XDM schema composition: classes, field groups, data types; identity descriptors and primary identity.
- Identity graph: how identities are stitched; namespace priority and the merge policy (default vs custom, timestamp-ordered vs dataset precedence).
- Segment builder: events (with time windows), profile attributes, and the PQL-style logic; sequential events and exclude logic.
- Dataset types: record vs time-series (experience events); data ingestion via streaming, batch, or source connectors; data prep mapping.
- Edge vs hub profile evaluation and segment evaluation methods (edge, streaming, batch).
Traps
- Assuming a segment immediately reflects a newly ingested attribute - merge policy and streaming vs batch evaluation change timing.
- Mixing up identity namespace conflicts: if two identities are not linked in the identity graph, the profile may split.
- Forgetting that time-series/experience event datasets require timestamp fields for event-based segmentation windows.
Worked rule of thumb
For "customer purchased in the last 30 days and is Gold tier," the event condition uses the experience-event dataset with a 30-day rolling window, and the attribute condition uses the profile record. If the question asks when the segment updates, check whether it's streaming (near-real-time) or batch (scheduled). When a profile seems missing data, inspect the merge policy and identity graph first.