Offer Decisioning allows you to use an trained model system that ranks offers to display for a given profile.
The use of AI ranking in Offer Decisioning is currently available in early access to select users only.
This feature enables you to create different ranking strategies based on your business goals. Using these different goal-based strategies in a decision (formerly known as offer activity), the trained model system will help you understand how the different ranking strategies are impacting your goals.
For example, you can select a ranking strategy for the email channel and another one for the push channel. For each channel, the trained model system will leverage multiple data points to determine which offer should be presented first for a given placement, rather than taking into account the offers’ priority scores or a ranking formula.
Once a ranking strategy has been created, assign it to a placement in a decision. Learn more in Configure offers selection in decisions.
To create a ranking strategy, follow the steps below:
Access the Components menu, then select the AI rankings tab.
All the ranking strategies created so far are listed.
Click the Create strategy button.
Fill in the following fields:
Name: Unique name that you must provide.
Model type: Currently the only supported model type is Auto-optimization.
This option enables marketers to choose how the machine-learn model should be built and trained: based on offers displayed, offers clicked in email, and/or offers clicked on the web.
You can select all metric types if needed.
There are two types of optimization metrics:
All selected impression events and/or conversion events will be automatically captured using the Web SDK or the Mobile SDK that has been provided. Learn more on this in Adobe Experience Platform Web SDK overview.
Dataset ID: For conversion, you need to provide a dataset where events are collected by selecting it from the drop-down list. Learn how to create such dataset in this section.
Only the datasets created from schemas associated with the Experience Event - Proposition Interactions field group (previously known as mixin) are displayed in the drop-down list.
Save and activate the ranking strategy.
It is now ready to be used in a decision to rank eligible offers for a placement. Learn more in this section.
You need to create a dataset where conversion events will be collected. Start by creating the schema that will be used in your dataset:
From the Data Management menu, select Schema, go to the Browse tab and click Create schema.
Choose XDM ExperienceEvent.
Learn more on XDM schemas and fields groups in the XDM System overview documentation.
In the Search field, type “proposition interaction” and select the Experience Event - Proposition Interactions field group.
The schema that will be used in your dataset must have the Experience Event - Proposition Interactions field group associated with it. Otherwise you will not be able to use it in your ranking strategy.
Click Add field groups.
Field group was previously known as mixin.
Type a name and save the schema.
Learn more on building schemas in Basics of schema composition.
You’re now ready to create a dataset using this schema. To do this, follow the steps below:
From the Data Management menu, select Datasets, go to the Browse tab and click Create dataset.
Select Create dataset from schema.
Select the schema you just created from the list.
Provide a unique name for the dataset in the Name field and click Finish.
The dataset is now ready to be selected to collect conversion events when creating a ranking strategy.