Skip to main content

Core Concepts & Terminology

Profile & Customer 360​

Everything CDP knows about a customer — a composite from multiple sources, not a single record.

Identity Resolution​

  • Deterministic — exact identifier matches, 100% confidence
  • Probabilistic — ML-based, requires thresholds
  • Unknown-to-known — linking anonymous behavior to identified customers

Traits & Scores​

Traits: SQL-computed attributes (total_spend_30d, engagement_score). Scores: ML predictions (churn_probability_30d, ltv_12m).

Segments​

Customer groups with a definition (criteria) and membership list (DTX_IDs). Created via visual builder, SQL, Python, or GenAI.

Events​

Timestamped records: track (actions), identify (updates), page/screen (navigation). Default 90-day retention.

Pipelines​

Source → processing → destination. Streaming (Flink), Batch (Spark), Hybrid.

Operators​

45+ building blocks across 10 categories. Some system-controlled, others user-editable.