You are here:
Configure an Objective-Based Recommender
Using a deep learning model, an objective-based recommender generates personalized, targeted recommendations for individuals to advance a specific business-related goal. You can choose to use an existing, predefined objective or create a new, custom objective for the recommender. Salesforce Personalization supports three predefined objectives–Maximize Revenue, Maximize Revenue with Promotions, and Maximize Clicks. These predefined objectives require you to map and include certain Data 360 entities in your profile and item data graphs. If you’re using a predefined objective, make sure you configure the data graphs correctly.
Required Editions
| User Permissions Needed | |
|---|---|
| To configure recommenders: | Change and Edit Recommenders |
- From the App Launcher, search for and select Recommenders.
- Click New.
-
In the Recommender Properties window, select a data space.
The data space that you choose determines which data graphs are available for you to select.
-
Select a profile data graph.
The profile data graph that you choose determines which profiles are eligible to receive personalized content and which profile-related attributes the personalization decision process can use. The menu lists only the profile data graphs in the selected data space.
Note f you want to use this profile data graph with a custom objective, make sure that it includes the engagement signals' DMOs and fields so that the signals are available for selection during objective creation. For example, the primary Data Model Object (DMO) should include parameters such as user ID, timestamp ID, and unique ID. Additionally, any DMOs and fields used in filters must be present and checked in the profile data graph. -
Select an item data graph to use for personalization response training.
The item data graph returns item-specific or catalog-specific content based on the data graph root object. The recommender uses this content when selecting what to return to the user in a personalized response. The menu lists only the item data graphs in the selected data space.Note If you want to use this item data graph with a custom objective, ensure that it includes the DMO and field used for the engagement signal's item identifier parameter. This will make the engagement signal available for selection during objective creation.
-
Enter a recommender name and an optional description.
The Recommender API Name is autogenerated.
- (Optional) Activate a fallback recommender. You can activate a fallback recommender at any time. See Set Up a Fallback Recommender.
- Click Next.
-
Select Objective-Based Recommendations, and then click Next. For
predefined objectives to function correctly, first configure the profile and item data
graphs with Data 360 entities.
- To use an existing or predefined objective, choose Existing
Objective and select an objective from the list.
objective description Maximize Revenue This objective returns recommendations intended to increase checkout frequency, increase cart value at checkout, or both. Make sure that the data graphs are configured correctly. Maximize Revenue with Promotions This objective returns promotion recommendations to influence product purchases through increased checkout frequency, increased cart value at checkout, or both.
By analyzing purchase history and order details across your website, app, and point of sale system, Salesforce Personalization identifies the promotions that actually trigger purchases. It then suggests the promotions that are most likely to increase total revenue based on each customer's unique shopping habits.
To effectively use this predefined objective, make sure that the event streams, data model, and data graphs are configured as outlined in the following topics.
Maximize Clicks This objective returns articles that are organized to maximize clicks based on current and historical article interaction from machine learning. Make sure that the data graphs are configured correctly. - To create a new objective, choose New Objective and use the instructions in Create a Custom Objective.
- To use an existing or predefined objective, choose Existing
Objective and select an objective from the list.
-
(Optional) Add filters.
You can add filters to a recommender at any time. See Add Recommendation Filters.
- Save the recommender.
Personalization creates the recommender. You can select the recommender for any personalization point decision.
- Create a Custom Objective
Using specific engagement signals, you can create custom business objectives to use with objective-based recommenders. You can reuse this objective to enhance the customer journey and deliver real-time, personalized recommendations, simplifying setup and boosting customer engagement. - Custom Objective for Offers
Custom objectives use deep learning models to predict the offers that are most likely to drive your business outcomes, even when customers don't directly interact with the offer itself.
