Cross Device Identity Management (CDIM) combines deterministic matching with probabilistic matching to link users across multiple cookies and devices. CDIM bridges a user's disparate ID’s into a single universal ID, which represents the user regardless of the device they are on.
What is the difference between probabilistic and deterministic matching?
Deterministic matching uses declarative data, meaning we know with certainty the device belongs to a particular user because the user has either logged-in or completed a registration process to declare his/her identity in association with that device.
Probabilistic matching, often called predictive modeling, relies on a collection of data that can be used to reliably determine a user identity across devices through statistical modeling. Implement probabilistic CDIM to extend the reach of your segments, and subsequently target these devices and people.
Important note: Probabilistic CDIM is a paid feature, which may not be part of your current contract. Please reach out to your account owner if you have questions about enabling this feature.
How to Build Segments Using CDIM
Choose the Base Segment
To create a probabilistic CDIM segment, first you need to create a base segment with the desired attributes and rules, or choose one from your existing standard segments.
Please note, If you are creating a new base segment from which you want to leverage the CDIM feature, the CDIM graph will not be available until the base segment has gone through the weekend processing queue. For example, if you create a new standard segment on Thursday, the CDIM graph should be available on the following Monday. Once the CDIM graph is available, you can proceed with building your CDIM segment.
Build the CDIM segment
- Find the segment you want to leverage and open the CDIM graph by clicking the action wheel beside the segment and selecting CDIM.
- The predictive cross-device graph will display, showing a bar for each level of similarity broken down by device type (i.e. Browser, iOS Mobile, or Android Mobile). Once you click the bar with the desired level of similarity, you will have the option to create the segment. (Level 5 is recommended for the most reach.)
- Select Create Segment, which will open the Segment Builder. The default setup in Segment Builder will use the original segment as the base, combining it with the predictive component using an OR operator. From this screen, you can also modify the rules of the segment before finalizing and saving. Note: You can delete the original base segment from the rules if desired.
- Complete the segment building process and save your CDIM segment. The segment will go into the processing queue and will populate within a few hours.
Tip: In the manage segments view, you can easily locate your CDIM segments using the filter:
Alternative CDIM Use Case: Create a segment of website visitors and app users
In addition to using CDIM to expand the reach of a segment, you may want to further understand the overlap of website visitors with app users. Using a standard non-CDIM segment will not work in this scenario because web visitors are tracked via cookies whereas app users would be tracked via Android IDs (ADIDs) and iOS IDs (IDFAs). The intersection of those will always be zero.
In this alternative scenario, CDIM can be used to allow for user matching to occur between desktop and apps. In the steps below, please note the correct rule set-up as the CDIM segment must be used as the base of the segment and additional rules must be nested.
Steps to Create the CDIM Segment of Website Visitors and App Users
- Create base segment of website visitors using Site Hierarchy. Allow the CDIM graph to populate for the segment.
- In Segment Insights, locate the segment and select Actions > Cross Device to view the predictive cross device graph once it is available.
- On the predictive cross device graph, select the desired scale of similarity, Select Create Segment.
- Set up the rules. The default setup will use the original segment as the base. This should be deleted to move the predictive CDIM segment to the base position. From there, additional rules from site hierarchy can be added as needed.
This setup assumes that Android and iOS users are captured within the CDIM segment and have been assigned uber IDs, which allows the overlap with the website to be identified using and ‘AND’ rule operator.