The pattern that adds context (weather, location, device state, session attributes, time-of-day, prior session memory) at the moment of decisioning rather than in batch. Required for recipes where the relevance of the decision depends on conditions changing faster than batch update cycles. Trades latency against richness, with edge-side enrichment forcing simpler downstream model architectures.
The pattern that enriches first-party data with external sources (firmographics, demographic data, intent providers, weather, market data) while tracking where each enriched field came from, when it was joined, and under what license. Provenance tracking is the discipline that makes third-party data removable when contracts end or accuracy degrades. Required for any defensible audit posture.
The pattern for derived signals computed from observed behavior rather than directly observed: lifetime value buckets, propensity scores, lifecycle stage, predicted segment membership. Distinct from raw data because inferences age, drift, and need refresh cycles. Distinct from real-time decisioning because the inferences are persistent attributes available to many downstream recipes.
The pattern that describes how a system remembers what it knows about a person across sessions and how that memory decays or refreshes. Increasingly important for AI-orchestrated experiences where the system needs to recall prior context to behave consistently. Distinct from raw event history because memory implies an editorial layer: what's worth remembering, what's worth surfacing, what should fade.
The pattern that triggers third-party or computed enrichment when a specific event fires, rather than enriching every record continuously. Used when enrichment costs (API calls, computation, vendor pricing) make per-event a wasteful default, but where the enrichment becomes load-bearing for downstream decisions at specific moments. Distinct from real-time contextual enrichment at the edge because the trigger is event-driven and the result is persistent, not request-scoped.
The pattern that loads a customer's relevant history into a decision at the moment the decision is made: pre-conversion events backfilled into a newly-stitched identity, full activity history attached to a sales handoff payload, prior session context surfaced to a personalization rule. Distinct from memory and recall because rehydration is the at-decision read pattern rather than the editorial layer that decides what to remember.