How Audience Analysis Works in CJA with RT-CDP ?

The Problem Statement

How do we connect RT-CDP audience segments with CJA event data to analyze customer journeys, despite differences in identity stitching models and delayed audience data availability?

Answer-

If you are setting up Adobe Customer Journey Analytics (CJA) along with Real-Time CDP (RT-CDP), you might notice an interesting challenge.

Both platforms handle identity stitching in very different ways:

• RT-CDP builds a real-time Identity Graph to link users across systems.
• CJA works on fast, query-based analysis using flattened event data (like web/mobile logs).

So the key question is:
How do these two systems connect?
How does an audience created in RT-CDP match perfectly with web or mobile events in CJA?

Let’s break it down step by step.

Core Concept: Persistent ID vs Person ID

When you configure a dataset (WebSDK or Mobile) in CJA, you define two important fields:

Persistent ID  →  Identity Graph  →  Person ID
(Raw Input)        (Linking)         (Final Output)

• Persistent ID (Input):
  This is the ID present in raw data (like ECID / device ID).
  It tracks anonymous users.

• Person ID (Output):
  This is the final unified ID (like Membership ID) used in CJA reports.

Step 1: How Stitching Works (Real-Time Flow)

When Graph-Based Stitching is enabled:

1. A user visits the website → event comes with ECID (example: Cookie_12345)
2. CJA queries AEP Identity Graph:
   “Is this ECID linked to any known user?”
3. Result:
   • If linked → Person ID = Membership ID
   • If not → Person ID = ECID (temporary fallback)

🔁 Replay Window (Important)

If the user logs in later (say Day 5):

• AEP links ECID → Membership ID
• CJA goes back (up to 14 days)
• Updates old anonymous events with the correct Person ID

✅ Result: Past anonymous activity becomes part of the unified profile

Step 2: Audience Mapping (CJA + RT-CDP Connection)

Even after stitching, CJA still doesn’t know audience segments.
That comes from RT-CDP.

To solve this, Adobe creates 2 datasets:

1. Profile Snapshot Dataset

• Contains all customer profiles
• Includes segment membership (audiences each user belongs to)

2. Audience Lookup Dataset

• Contains:
  • Audience ID
  • Audience Name
  • Description

🔗 How Mapping Works

Event Data (Person ID)
        ↓
Match with Profile Dataset
        ↓
Check Segment Membership
        ↓
Use Lookup Dataset to show Audience Name

Example:

• Person ID = Membership_999
• Profile shows audience ID
• Lookup converts it to: "Non Loyalty Members"

✅ Now CJA can show audience data in reports

🔹 Important Limitations / Gotchas

1. Not Real-Time (Audience Delay)

• Profile snapshot is updated once daily
• Audience data in CJA = yesterday’s view

2. Streaming Audience Issue

• Streaming audiences update instantly in AEP
• But they won’t appear in CJA unless batch evaluation is enabled

3. Data Retention in CJA

• If CJA keeps only 30 days of data:
  • You will see only users active in last 30 days
  • Even if audience size is larger in AEP

✅ Final Summary

• Identity Graph (AEP): Links device IDs to real users
• Profile Service (AEP): Builds audiences
• CJA:
  • Uses stitched Person ID
  • Matches with profile data
  • Adds audience details in reports

👉 In short:
AEP handles identity + segmentation, CJA uses it for analytics and reporting.