Build a cross-channel analysis workspace
Stitch behavioral data from web, voice, and call-center channels into a unified person-level connection, define a data view with dimensions and metrics, and surface journey funnels and flow visualizations in an analysis workspace.
A cross-channel analysis workspace unifies behavioral signals from every customer touchpoint — web sessions, voice interactions, call-center transcripts — into a single person-level view that makes previously invisible journey segments visible to analysts without requiring SQL skills. The workspace is built in two layers: a data connection that stitches raw channel datasets on a shared identity key, and a data view that maps raw field paths to analyst-friendly dimension and metric names.
The stitching decision is the most consequential configuration choice. The Person ID field must exist in every dataset being joined; using email address is common but introduces latency if a visitor is not yet authenticated. Practitioners must also decide on backfill depth, session timeout thresholds, and how to handle datasets that lack a direct Person ID (using a derived field or a lookup dataset). Once the connection is live, the data view exposes the joined schema to the workspace layer where funnels, flows, and attribution analyses are assembled as reusable components.
Parallel viability: Medium parallelism. Snowflake or Databricks with a BI front-end (Looker, Tableau, Metabase) can replicate the cross-channel join and funnel visualization patterns. The primary CJA advantages are real-time data freshness from the AEP streaming pipeline, native session-level stitching that uses AEP's Identity Service graph, and the no-code workspace experience aimed at marketing analysts. Teams with existing data-warehouse infrastructure and analyst SQL fluency should evaluate whether the composable path meets their latency and identity-resolution requirements before adopting CJA.
Side-by-side implementations
In Customer Journey Analytics (CJA), practitioners create a Connection by selecting AEP datasets (Website Interactions, Voice Assistants, Call Center interactions) and designating a common Person ID field (e.g., email address from the AEP Profile dataset) that stitches records across sources. A Data View is built on top of the Connection, mapping XDM fields to named dimensions (Page Name, Phone Number) and metrics (Page Views), configuring persistence and attribution settings per component. Analysis Workspace projects are then assembled with freeform tables (e.g., top products by page views), fallout visualizations tracking the conversion sequence from productViews through productListAdds to purchases, and flow visualizations that reveal paths following key events such as a "Cancel Service" interaction traced into call-center follow-up actions.
Capability: Identity Resolution
In a Snowflake-native analytics stack, the equivalent of CJA's Connection is a set of JOIN keys declared in a dbt model that unifies event streams from web, call-center, and CRM tables under a shared person_id from the identity map table. The equivalent of a CJA Data View is a semantic layer definition in dbt Metrics (dbt 1.6+) or a Snowflake Semantic View (Snowflake Cortex Analyst preview) that maps physical columns to named dimensions and metrics with persistence and attribution settings. Analysis Workspace projects are replaced by Tableau, Looker, or Mode workbooks connected via JDBC/ODBC to Snowflake. Cross-channel freeform tables, fallout funnels, and flow visualizations are native chart types in all major BI tools. The key difference from CJA is that identity stitching is done via a pre-computed identity map table refreshed by the identity-stitching dbt model, rather than at query time using AEP's Identity Service graph.
Capability: Identity Resolution
Hightouch does not provide a CJA-equivalent cross-channel analysis workspace — there is no Connection, Data View, or Analysis Workspace analogue in Hightouch. Ad-hoc cross-channel analysis and funnel visualization remain in the Snowflake + dbt + BI tooling layer documented in Phase 3. Hightouch's role in the composable equivalent is as the activation bridge: once the Snowflake/BI analysis identifies a meaningful cross-channel cohort (e.g., customers who had a "Cancel Service" call-center interaction within 7 days of a product-page visit), that cohort is expressed as a Hightouch Audience filter against the dbt identity-map model and activated to a retention channel — closing the insight-to-action loop that CJA provides natively via the shared AEP data layer.
Capability: Audience Segmentation
Task-level sources
- technical-training/module11/index.md
- technical-training/module11/ex1.md
- technical-training/module11/ex2.md
- technical-training/module11/ex3.md
- technical-training/module11/ex5.md
- technical-training/module11/summary.md
How is this implementation?
Sign-in-gated. Tomorrow morning's curriculum-ingestor consumes your feedback: "Inaccurate" queues the task for re-review, "needs update" queues it for a refresh, and "one vendor panel is wrong" re-drafts just that panel.