Adobe Stack Daily #006 — CJA and AEP Are Showing Different Numbers. Both Are Correct. Here’s Why.

2026-05-05 · Source: Data Engineering on Medium · Field: Technology & Digital — Data Science & Analytics, Cloud Computing & IT Infrastructure · Depth: Advanced, long

Summary

Adobe Stack Daily #006 addresses common discrepancies between Adobe Customer Journey Analytics (CJA) and Adobe Experience Platform (AEP) Query Service, where the same underlying data can yield vastly different metric counts. This occurs because AEP functions as a raw data platform, storing events as ingested, while CJA is an analytics layer that applies transformations like identity stitching, sessionization, attribution, and deduplication at query time. The article details four architectural reasons for these differences: CJA's person deduplication versus AEP's raw identity counts, CJA's configurable session logic versus AEP's lack of session concept, CJA's attribution and metric deduplication versus AEP's raw event storage, and CJA's exclusion of rows missing a Person ID. It also highlights data freshness gaps due to CJA's processing latency.

Key takeaway

For data analysts and engineers comparing metrics across Adobe CJA and AEP, you must understand the four architectural differences detailed. Do not assume discrepancies are bugs; instead, document CJA's Person ID configuration, stitching method, session definitions, and metric deduplication settings. This proactive approach will prevent stakeholder confusion and ensure clarity on what each system's numbers truly represent, avoiding misinterpretations of data quality.

Key insights

CJA and AEP yield different metrics from the same data due to distinct processing logic and architectural roles.

Principles

• AEP stores raw events; CJA applies transformations.
• Different systems measure different aspects of data.
• Document system logic before comparing metrics.

Method

To reconcile CJA and AEP discrepancies, define Person ID, confirm stitching, understand session definitions, and verify metric deduplication settings in CJA Data Views.

In practice

• Use AEP Audience Portal for audience sizing.
• Validate CJA Connection skip rates per dataset.
• Configure deduplication for all CJA conversion metrics.

Topics

• Adobe Customer Journey Analytics
• Adobe Experience Platform
• Data Discrepancy Resolution
• Identity Stitching
• Session Definition

Best for: Data Scientist, Data Engineer, Analytics Engineer

Related on AIssential

• Java News Roundup: OpenJDK JEPs, Jakarta EE 12, Spring Framework, Micrometer, Camel, JBang — Key Java ecosystem components are advancing with new features, performance enhancements, and critical security fixes.
• Article: Jakarta EE 12 Milestone 2: Advent of the Data Age Along with Consistency and Configuration — Jakarta EE 12 unifies data querying, modernizes Java baselines, and integrates agent-based AI capabilities.
• Pretrained, Frozen, Still Leaking: Auditing Cross-Encoder Attribute Transfer in EEG Foundation Models — Single-endpoint audits for EEG foundation models are insufficient, as spectral attributes can still leak via cross-encoder transfer.
• How to Measure Token Impact of MCP Tool Invocation in Microsoft Foundry — Token accounting for MCP tool invocation in Microsoft Foundry requires disciplined evidence handling across disparate telemetry sources.
• Benchmarking RAG Architectures Locally on a Real Financial PDF — Part 3: The Measurement Problem — Multimodal RAG evaluation requires answer correctness against independent references, as text-grounded metrics misrepresent performance.
• Description-Code Inconsistency in Real-world MCP Servers: Measurement, Detection, and Security Implications — Semantic mismatches between LLM tool descriptions and code, termed DCI, pose critical reliability and security threats.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Data Engineering on Medium.