Journey IQ: Cross-Channel Analytics is a feature that allows you to re-key a dataset’s person ID, which enables a seamless combination of multiple datasets. CCA looks at user data from both authenticated and unauthenticated sessions to generate a stitched ID. Using Cross-Channel Analytics, you can answer questions such as:
How many people begin their experience in one channel, then finish it in another?
How many people interact with my brand? How many and what types of devices do they use? How do they overlap?
How often do people begin a task on a mobile device and then later move to a desktop PC to complete the task? Do campaign click-throughs that land on one device lead to conversion somewhere else?
How does my understanding of campaign effectiveness change if I take into account cross-device journeys? How does my funnel analysis change?
What are the most common paths users take from one device to another? Where do they drop out? Where do they succeed?
How does the behavior of users with multiple devices differ from the users with a single device?
When you combine datasets with similar person IDs, attribution is carried over across devices and channels. For example, a user first visits your site through an advertisement on their desktop computer. That user encounters an issue with their order, then gives your customer service team a call to help resolve it. With Cross-Channel Analytics, you can attribute call center events to the ad that they originally clicked.
Failure to meet all prerequisites can result in the inability to create a CCA connection or poor results when combining datasets.
Before using Cross-Channel Analytics, make sure that your organization is prepared with the following:
One dataset in Adobe Experience Platform must have two columns that help identify visitors:
A persistent ID, an identifier present on every row. For example, a visitor ID generated by an Adobe Analytics AppMeasurement library.
A transient ID, an identifier present on only some rows. For example, a hashed username or email address once a visitor authenticates. You can use virtually any identifier that you’d like, as long as it is present at least once on the same event as a given persistent ID.
Another dataset, such as call center data, that contains a transient ID on every row. This person ID must be formatted similarly as the transient ID in the other dataset.
This feature allows you to stitch together datasets which can include merging authenticated and unauthenticated user data. Please ensure that you comply with applicable laws and regulations, including obtaining necessary end-user permissions, before merging datasets.
Any change to the global event dataset schema has to be applied also in the new stitched dataset schema, otherwise it will break the stitched dataset.
Also, if you remove the source dataset, the stitched dataset stops processing and gets removed by the system.
Cross-Channel Analytics is a groundbreaking and robust feature, but has limitations on how it can be used.
Current rekeying capabilities are limited to one step (persistent ID to transient ID). Multiple-step rekeying (for example, persistent ID to a transient ID, then to another transient ID) is not supported.
Only event datasets are supported. Other datasets, such as lookup datasets, are not supported.
Custom ID maps used in your organization are not supported.
The Cross-Device Private graph is not supported.
Cross-Channel Analytics does not transform the field used for stitching in any manner. Field-based stitching uses the value in the specified field as it exists in the unstitched dataset within data lake. The stiching process is case sensitive. For example, if sometimes the word ‘Bob’ appears in the field, and sometimes the word ‘BOB’ appears, these will be treated as two separate people.
Given field-based stitching is case-sensitive, for Analytics datasets generated through the Analytics Source Connector, Adobe recommends reviewing any VISTA rules or processing rules that apply to the transient ID field to ensure that none of these rules are introducing new forms of the same ID. For example, you should ensure that no VISTA or processing rules are introducing lowercasing to the transient ID field on only a portion of the events.
Field-based stitching does not combine or concatenate fields.
The transient ID field should contain a single type of ID (i.e. IDs from a single namespace). For instance, the transient ID field should not contain a combination of login IDs and email IDs.
If multiple events occur with the same timestamp for the same persistent ID, but with different values in the transient ID field, field-based stitching will choose based on alphabetical order. So if persistent ID A has two events with the same timestamp and one of the events specifies Bob and the other specifies Ann, field-based stitching will choose Ann.
If a device is shared by multiple people and the total number of transitions between users exceeds 50.000, CCA stops stitching data for that device.
Enable Cross-Channel Analytics
Once your organization meets all prerequisites and understands its limitations, you can follow these steps to start using it in CJA.
Contact Adobe Customer Support with the following information:
A request to enable Cross-Channel Analytics
The dataset ID for the dataset that you want to rekey
The column name of the persistent ID for the desired dataset (Identifier that appears on every row)
The column name of the transient ID for desired dataset (The person identifier link between datasets)
Your preference of replay frequency and lookback length. Options include a replay once a week with a 7-day lookback window, or a replay every day with a 1-day lookback window
The Adobe Customer Support will work with Adobe engineering to enable Cross-Channel Analytics upon receiving your request. Once enabled, a new rekeyed dataset that contains a new person ID column appears in Adobe Experience Platform. Adobe Customer Support can provide the new dataset ID and person ID column name.
When first turned on, Adobe will provide a backfill of stitched data that goes back as far as the beginning of the previous month (up to 60 days.) In order to do this backfill, the transient ID must exist in the unstitched data back that far in time.
Create a connection in CJA using the newly generated dataset and any other datasets that you want to include. Choose the correct person ID for each dataset.