蜜豆视频

DocumentationExperience PlatformPlatform Identity Service Guide

Identity graph linking rules overview identity-graph-linking-rules-overview

Last update: Thu Mar 20 2025 00:00:00 GMT+0000 (Coordinated Universal Time)

CREATED FOR:

  • Admin
  • Developer
AVAILABILITY
Identity graph linking rules are currently in Limited Availability. Contact your 蜜豆视频 account team for information on how to access the feature in development sandboxes.

With 蜜豆视频 Experience Platform Identity Service and Real-Time Customer Profile, it is easy to assume that your data is ingested perfectly and that all merged profiles represent a single individual person through a person identifier, such as a CRMID. However, there are possible scenarios where certain data could try to merge multiple disparate profiles into a single profile (鈥済raph collapse鈥). To prevent these unwanted merges, you can use configurations provided through identity graph linking rules and allow for accurate personalization for your users.

Watch the following video for additional information on using identity graph linking rules:

video poster

Transcript
Hi, I鈥檓 excited to give you this quick overview of Identity Graph Linking Rules in 蜜豆视频 Experience Platform. This feature helps data architects build quality identity graphs and customer profiles for 蜜豆视频 Experience Platform applications like Real-Time Customer Data Platform, Journey Optimizer, and Customer Journey Analytics. It helps avoid graph collapse, which is when the identity graphs of two people unintentionally merge together. When people engage with your brand, data flows into 蜜豆视频 Experience Platform datasets. If those datasets are enabled for profile, identity graphs form and profile fragments are stored. If two individual people end up sending data with common identities, their identity graphs files can merge and give you an inaccurate representation of your customers. This can easily happen when people share devices, enter fake information, or through unintentional implementation errors. Let鈥檚 take a look at the interface. From the left navigation, select Identities. There are two components of the feature, Graph Simulation and Identity Settings. It can be tested and configured entirely in the interface and doesn鈥檛 require any code changes elsewhere. To fully use it, you need permissions to view and manage identities. I鈥檒l start in the Graph Simulator. The simulator has several examples you can load to get started, which correspond to common scenarios that could lead to graph collapse. Let鈥檚 take a look at the shared device example. At the top of the screen, we have Events, and at the bottom is the namespace configuration. Events represent any data ingestion event containing two or more identities, which is the minimum required for graph construction. These can be XDM Experience events or ingestion events containing XDM individual profile data. You can add events using a UI dialog or paste text events using the Text option. Events just need to contain pairs of identity namespaces and values. The algorithm configuration is the heart of the identity graph linking rules. You can prioritize namespaces by dragging and dropping, and declare uniqueness by checking the box. The priority impacts the order of rule evaluation, and uniqueness means that only one value for the namespace is allowed per graph. Hit the simulate button to see how your configuration will play out on the events. Here you can see that although John and Jane shared a device and had a common ECID, they end up with separate identity graphs because the hashed email identity was declared unique, and the rules removed the link between them. Without configuring identity graph linking rules or without using the unique setting, their graphs and profiles would merge, which I can simulate by unchecking the uniqueness box and hitting the simulate button again. The graph simulation is a great tool to test out complex scenarios using events and namespaces representative of your own implementation. Once you鈥檝e determined the priority and uniqueness settings you need for your implementation, you can set up the final configuration on the settings screen, which you can get to by clicking on the settings button. This screen is similar to the namespace configuration portion of the simulator. It pulls in all of the identity namespaces in your sandbox, and you can change the priority and uniqueness settings. The highest priority namespace must be unique. Once you activate your rules, your existing graphs will remain, but will be recreated as graphs are updated by new events ingested into platform. That鈥檚 a really quick overview. It鈥檚 a simple feature to configure, but because identity graphs resonate throughout so many services in Experience Platform, especially the profile service, it鈥檚 important to configure it with consideration. Thank you.

Get started

The following documents are essential in understanding identity graph linking rules.

Graph collapse scenarios graph-collapse-scenarios

This section outlines example scenarios that you may consider when configuring identity graph linking rules.

Shared device

There are instances where multiple logins can occur on a single device:

Shared device
Description
Family computers and tablets
Husband and wife both login to their respective bank accounts.
Public kiosk
Travelers at an airport logging on using their loyalty ID to check in bags and print boarding passes.
Call center
Call center personnel log in on a single device on behalf of customers calling customer support to resolve issues.

A diagram of some common shared devices.

In these cases, from a graph standpoint, with no limits enabled, a single ECID will be linked to multiple CRMIDs.

With identity graph linking rules, you can:

  • Configure the ID used for login as unique identifier. For example, you can limit a graph to store just one identity with a CRMID namespace, and thus define that CRMID as the unique identifier of a shared device.
    • By doing this, you can ensure that CRMIDs do not get merged by the ECID.

Invalid email/phone scenarios

There are also instances of users who provide fake values as phone numbers and/or email addresses when registering. In these cases, if limits are not enabled, then phone/email related identities will end up being linked to multiple different CRMIDs.

A diagram that represents invalid email or phone scenarios.

With identity graph linking rules, you can:

  • Configure either the CRMID, phone number, or email address as the unique identifier and thus limit one person to just one CRMID, phone number, and/or email address associated with their account.

Erroneous or bad identity values

There are cases where non-unique, erroneous identity values are ingested in the system, irrespective of namespace. Examples include:

  • IDFA namespace with the identity value of 鈥渦ser_null鈥.
    • IDFA identity values should have 36 characters: 32 alphanumeric characters and four hyphens.
  • Phone number namespace with the identity value of 鈥渘ot-specified鈥.
    • Phone numbers should not have any alphabet characters.

These identities could result in the following graphs, where multiple CRMIDs are merged together with the 鈥榖ad鈥 identity:

A graph example of identity data with erroneous or bad identity values.

With identity graph linking rules you can configure the CRMID as the unique identifier to prevent unwanted profile collapsing due to this type of data.

Identity graph linking rules identity-graph-linking-rules

With Identity graph linking rules you can:

  • Create a single identity graph / merged profile for each user by configuring unique namespaces, which will prevent two disparate person identifiers from merging into one identity graph.
  • Associate online, authenticated events to the person by configuring priorities

Terminology terminology

Terminology
Description
Unique namespace
A unique namespace is an identity namespace that has been set up to be distinct within the context of an identity graph. You can configure a namespace to be unique using the UI. Once a namespace is defined as unique, a graph can only have one identity that contains that namespace.
Namespace priority
Namespace priority refers to the relative importance of namespaces compared to one another. Namespace priority is configurable through the UI. You can rank namespaces in a given identity graph. Once enabled, names priority will be used in various scenarios, such as input for identity optimization algorithm and determining primary identity for experience event fragments.
Identity optimization algorithm
The identity optimization algorithm ensures that guidelines created by configuring a unique namespace and namespace priorities are enforced in a given identity graph.

Unique namespace unique-namespace

You can configure a namespace to be unique using the identity settings UI workspace. Doing so, informs the identity optimization algorithm that a given graph may only have one identity that contains that unique namespace. This prevents the merging of two disparate person identifiers within the same graph.

Consider the following scenario:

  • Scott uses a tablet and opens his Google Chrome browser to go to acme.com, where he signs in and browses for new basketball shoes.

    • Behind the scenes, this scenario logs the following identities:

      • An ECID namespace and value to represent the use of the browser
      • A CRMID namespace and value to represent the authenticated user (Scott signed in with his username and password combination).
  • His son Peter then uses the same tablet and also uses Google Chrome to go to acme.com, where he signs in with his own account to browse for football equipment.

    • Behind the scenes, this scenario logs the following identities:

      • The same ECID namespace and value to represent the browser.
      • A new CRMID namespace and value to represent the authenticated user.

If CRMID was configured as a unique namespace, then the identity optimization algorithm splits the CRMIDs apart into two separate identity graphs, instead of merging them together.

If you do not configure a unique namespace, you may end up with unwanted graph merges, such as two identities with the same CRMID namespace, but different identity values (scenarios like these often represent two different person entities in the same graph).

You must configure a unique namespace to inform the identity optimization algorithm to enforce limitations on the identity data that are ingested into a given identity graph.

Namespace priority namespace-priority

Namespace priority refers to the relative importance of namespaces compared to one another. Namespace priority is configurable through the UI and you can rank namespaces in a given identity graph.

One way in which namespace priority is used is in determining the primary identity of experience event fragments (user behavior) in Real-Time Customer Profile. If priority settings are configured, then the primary identity setting on Web SDK will no longer be used to determine which profile fragments are stored.

Unique namespaces and namespace priorities are both configurable in the identity settings UI workspace. However, the effects of their configurations are different:

Identity Service
Real-Time Customer Profile
Unique namespace
In Identity Service, the identity optimization algorithm refers to unique namespaces to determine the identity data that is ingested to a given identity graph.
Unique namespaces do not affect Real-Time Customer Profile.
Namespace priority
In Identity Service, for graphs that have multiple layers, namespace priority will determine that the appropriate links are removed.
When an experience event is ingested in Profile, the namespace with the highest priority becomes the primary identity of the profile fragment.
  • Namespace priority does not affect graph behavior when the limit of 50 identities per graph is reached.
  • Namespace priority is a numerical value assigned to a namespace indicating its relative importance. This is a property of a namespace.
  • Primary identity is the identity in which a profile fragment is stored against. A profile fragment is a record of data that stores information about a certain user: attributes (usually ingested via CRM records) or events (usually ingested from experience events, or online data).
  • Namespace priority determines the primary identity for experience event fragments.
    • For profile records, you can use the schemas workspace in the Experience Platform UI to define identity fields, including the primary identity. Read the guide on defining identity fields in the UI for more information.
  • If an experience event has two or more identities of the highest namespace priority in the identityMap, it will be rejected from ingestion because it will be deemed as 鈥渂ad data鈥. For example, if the identityMap contains {ECID: 111, CRMID: John, CRMID: Jane}, the entire event will be rejected as bad data because it implies that the event is associated to both CRMID: John and CRMID: Jane simultaneously.

For more information, read the guide on namespace priority.

Next steps

For more information on identity graph linking rules, read the following documentation:

recommendation-more-help
64963e2a-9d60-4eec-9930-af5aa025f5ea