About visitor stitching
Visitor stitching is a form of identity resolution that works across platforms, devices, and browsers to associate activity to the same visitor. This article describes how visitor stitching works, how to use the visitor ID attribute effectively, and best practices.
Identifying your visitors is a critical step in making full use of Tealium AudienceStream CDP. When certain qualifying visitor events occur, such as a user registering or completing a purchase, there is an opportunity to uniquely identify that visitor using an event attribute such as email_address_hash
or customer_id
. When one of these identifying attributes is provided in the data layer across multiple sessions, AudienceStream can combine the visitor attributes from those sessions into one master visitor profile using a process called visitor stitching.
Contact your account manager to have visitor stitching enabled for your account and profile.
The Tealium identity resolution process uses several types of attributes, as follows:
- Anonymous identifier – an anonymous, unique value generated by Tealium Collect for each visit to a site or each use of an app.
- User identifier – an identifier, such as an email address, that uniquely identifies a user across devices, platforms, and browsers.
- Visitor ID attribute – an attribute that is used in visitor stitching to identify users across sessions and devices. A visitor ID attribute gets its value from a user ID and is used to combine combine visitor profiles that have the same visitor ID attribute value into a single profile. Selecting the visitor ID attributes you will use for visitor stitching is an important part of AudienceStream configuration. For more information, see Visitor ID Attributes.
For more information, see Anonymous IDs, User IDs, and Visitor ID Attributes.
How identity resolution works
Identity resolution is the process of collecting and matching user identifiers across browsers and devices to create a unified view of a visitor’s engagement with your brand.
For example, when a new visitor interacts with your brand, on your website or app, Tealium Collect generates an anonymous ID and includes it in the data sent to AudienceStream. This results in a new visitor profile, a record in AudienceStream that stores visitor attributes. The same visitor might have several profiles in AudienceStream, one for each device or browser, and over time they accumulate more data.
When a visitor validates their identity, by providing a user identifier such as an email address, Tealium Collect sends this user identifier to AudienceStream to indicate that it is a known visitor. The user identifier populates a visitor ID attribute in AudienceStream. Visitor stitching is triggered the first time a specific visitor ID attribute is populated from a user ID attribute in AudienceStream.
How profiles are stitched together
Visitor stitching goes beyond combining the data from two or more profiles into a single profile. To create the most accurate unified profile possible, Tealium’s proprietary technology replays the data from each of the stitched profiles, in order of occurrence, building the new profile based on the sequence of the visitor’s journey.
The stitched profiles are not deleted. The new profile contains a replaces
array that contains links to each stitched profile. Each of the stitched profiles contains a replaced by
attribute that contains a link to the new profile. The anonymous ID in a stitched profile can still be used to look up or delete a visitor. Events that reference an anonymous ID in a stitched profile will now enrich the new profile. For more information about replaces
attributes, see AudienceDB tables and Visitor record format.
Stitching occurs in real-time and newly combined profiles can be immediately acted upon, thus making personalization, remarketing, and retargeting actions much more personalized and effective.
For more information, see Visitor Stitching Example.
After visitor profiles have been stitched, they cannot be separated.
Visitor switching
When two users share a device, both users have the same anonymous ID. If they visit the same websites, but use different user identifiers (such as an email address or a customer ID), this is referred to as visitor switching. Tealium Support or Professional Services can assist you with implementing a solution to handle visitor switching. For more information, see the following Tealium knowledge base articles:
- Anonymous User Tracking under ITP with Visitor Switching in AudienceStream
- Visitor Switching in iQ
- Visitor Switching in Tealium Mobile SDKs
How AudienceStream evaluates identifiers
When AudienceStream processes events, the anonymous ID (tealium_visitor_id
) is evaluated before any user identifiers or visitor ID attributes. The only exception to this order of evaluation is that for events from file import data sources, the evaluation order is determined by the visitor ID mappings.
The following scenarios describe how anonymous IDs and user identifiers affect the visitor profiles in AudienceStream:
-
New Anonymous Visitor – If the anonymous ID does not match an existing profile, a new visitor profile is created.
-
Updated Anonymous Visitor – If the anonymous ID matches a visitor profile, the event enriches the existing profile.
-
Identified Visitor – If the event includes a user identifier that is configured to enrich a visitor ID attribute, and the visitor ID attribute is not populated already, the user identifier enriches the visitor ID attribute and the event enriches the existing profile.
-
Anonymous Visitor Matches Identified Visitor – If the event includes a user identifier that is configured to enrich a visitor ID attribute, and the user identifier matches the visitor ID attribute in an existing profile, the two profiles are stitched together.
-
Multiple Identifiers – If the event contains multiple user identifiers, the corresponding visitor ID attributes are evaluated as an OR condition in attribute ID order. This means that the visitor ID attribute that was created first is evaluated first. In some edge cases, the order of evaluation can affect how profiles are stitched, particularly when the profiles have no common anonymous IDs.
This page was last updated: June 12, 2024