Practitioner tool

Stitching Explorer

How CJA cross-device stitching resolves one person from many devices — step through live person-ID resolution at ingest, then replay the lookback window that reclaims the anonymous pre-login journey.

Step 1 of 2

Two algorithms

Elevating device IDs to person IDs

Both algorithms take an anonymous device ID and try to resolve it to a known person ID. They differ in where they look for the identity signal — and how they handle shared devices.

Field-Based Stitching

Identity sourceWithin a single dataset — finds a login event carrying the person ID in the same rows being stitched.

How it resolvesMaps device ID → person ID whenever a login event is found in that dataset.

ReplayLooks back over a configured window to re-attribute historical rows retroactively.

Shared devicesDevice-split — forwards credit to the most recent authenticated user in time.

Best forA single dataset with frequent logins; simpler setup; no identity graph required.

Simpler setup

Graph-Based Stitching

Identity sourceFull AEP Identity Graph — shared with Real-Time CDP and Journey Optimizer, spanning multiple datasets.

How it resolvesResolves identities across datasets via multi-hop graph relationships.

ReplaySame scheduled replay window; the graph is consulted at replay time for better resolution.

Shared devicesLast-auth — credits the last authenticated person, regardless of device.

Best forMulti-dataset, cross-channel scenarios — when the identity lives in the graph but not the dataset.

Recommended for cross-channel

A typical enterprise has web events (ECID), CDP profiles (person ID), email campaigns (hashed email) and call-centre records (phone) — four datasets, four different primary identities. Only graph-based stitching can bridge all four through the identity graph, connecting one person’s full cross-channel journey.