Einstein Decisions analyzes many unique data points when determining the right
experience to present to a particular user. This data includes user behaviors as well as
contextual data, such as device type, time of day, and demographic data. Einstein Decisions
can also reference segment membership and user attribute data, along with catalog objects you
create in your catalog. To determine which data Einstein Decisions references, you define the
machine learning features on the Feature Engineering page. Typically, a data science or
analytics team member configures the feature engineering options for Einstein
Decisions.
Required Editions
| Available in: Premium edition |
| Permissions
Needed |
| To configure Einstein Decisions: |
A role with Contextual Bandit Create/Edit permissions |
-
From the Machine Learning section of the main navigation,
select .
-
Select the training target from the Bandit Target
dropdown—Click, Conversion,
or Goal Completion.
-
Select the features you want to use while training the machine learning
models.
-
To include specific user segments, click +Segment in the
Segment Membership section and select options from the
dropdown.
-
To include specific locations, click +Geography in the
Geography section and select options from the
dropdown.
-
To include custom attributes, click +Attribute in the
Custom Attribute section and select options from the
dropdown.
-
To include recent user actions with catalog objects, click +Catalog
Object in the Catalog Object Interaction
History section and select options from the dropdown.
-
Click Save.
After configuring the training target and feature engineering options, you
can create campaigns with promotions that use Einstein Decisions.
