Can Target's machine-learning use Audiences created in the Target Audience Library? | Community
Skip to main content
K
kinggi
Level 2
May 13, 2026
Question

Can Target's machine-learning use Audiences created in the Target Audience Library?

  • May 13, 2026
  • 2 replies
  • 79 views

Can Auto-Target and Auto Personalisation’s machine-learning use audiences created in the Target Audience Library directly?

The documentation (What Data is Collected to Build Machine-Learning Algorithms? | Adobe Target) specifically references audiences created in Experience Cloud (with examples of Adobe Analytics and Adobe Audience Manager), as well as audiences created in Platform CDP, but it is not clear whether audiences created in Target’s audience library are used by the machine-learning.

For context: We have a single mbox parameter which contains a comma delimited string of identifiers which relate to audiences defined in an external platform. We create audiences in the Target Audience Library to split out each separate audience and make it usable. The string itself will be relatively useless to the ML models. Ideally the audiences we already create in the Target UI will be available.

Thanks!

2 replies

S
Adobe Employee
SarthakBhAdobe Employee
Adobe Employee
May 14, 2026

I would recommend not passing multiple values in the same attribute or variable. If multiple variables are concatenated, Target’s personalization algorithms treat each string as a unique value, reducing the value of the information for personalization, which reduces usefulness for learning.

So if you currently have something like:

audiences="segA,segB,segC"

i would do 

profile.ext_aud_segA 
profile.ext_aud_segB 
profile.ext_aud_segC 

 

Also i do belive a Target audience can absolutely be used for targeting / reporting setup.

 

Thank you! 

Sarthak Bhatt | Senior Product Consultant | Customer Experience | Adobe  

 

K
kinggiAuthor
Level 2
May 16, 2026

Thanks for your reply Sarthak. In our situation the concatenated string was the right approach. The question is specifically whether Target Audiences built using the Target UI and in the Audience Library are part of the Machine Learning algo data inputs. The documentation talks about certain types of audiences as being in scope but is silent on this specific type of audience. Do you know the answer?

-Steve

P
PrasanthV
Level 3
May 20, 2026

Hi ​@kinggi ,

 

The Short Answer: No, but with a major catch

The machine-learning models cannot directly read or ingest a "Target Audience Library" rule as a feature vector.

When the documentation states that Adobe Analytics, Audience Manager, or Real-Time CDP audiences are used, it is because those platforms pass predefined, static Segment IDs straight into the visitor profile before the evaluation happens.

Terms & ConditionsAccessibility statement

Loading footer...