Overview
A Dynamic Recommendation consists of items from a Product Catalog that are grouped based on actions performed by users. Using Recommendations, you can create campaigns that focus on items on which the user has performed a particular action such as adding items to a shopping cart and not completing the purchase or searching for a hotel and not completing the booking action.
In such cases, using Recommendations you can create a Cart Abandonment campaign that runs every time a user abandons their cart, and contains information about the items abandoned. What’s more, you can even personalize this email, include customized offers, and decide on specific conditions that need to be met for the email to be triggered.
library_add_check |
Prerequisites Refer to Product Catalogs and User Action Mapping before creating Recommendations. |
Creating Recommendations
Recommendations can be created from the Content > Recommendations menu option on the MoEngage platform. Click on this link to navigate to the Recommendation creation page. Since a Recommendation is a result of User Actions performed on items contained in Product Catalogs, to create a Recommendation, we need to specify the following information:
- Product Catalog – This tells the Dynamic Recommendation which Product Catalog contains the items that will be used in the email campaign (for example, Groceries, Hotels, All_Items, and so on). While there can be multiple product catalogs created, currently a Recommendation can be created on only one Product Catalog at a time.
- User Action – This is required to identify which user actions (for example, Added to Cart, Removed from Cart, and so on) need to be used to fetch items from the Product Catalog specified in the previous step. Users can use a combination of actions to precisely define the conditions for the email campaigns to be triggered.
There are the following Recommendation recipes available that can be used across different channels e.g. Push, SMS, Email, In-App, OSM, and Cards. Select any of these recipes while creating a Recommendation.
User Actions
Select the user actions and the period when the user actions are performed. Here user action provides the Items, on which the filters are applied. You can choose to add AND/OR and exclude filters to get the desired list of items for a user.
For example - Get all the items for a user, where the user has viewed the items (Product Viewed) and added those items to the cart (Added to Cart) and has not purchased those items (Exclude Product Purchased).
Item Attributes
You can filter the items based on the item attributes. You can choose to add AND/OR and exclude filters to get the desired list of items. The filter value for item attributes can be a custom value or it can also be a value that is present in any user attributes.
If filtering is based solely on a custom value, the resulting filtered items will be accessible to all users. This creates a generic item list that is the same for all users.
If items are filtered with at least one user attribute, the items are filtered for the specific user. Here the item list is specific/personalized to each user. If you want to personalize the items for users based on user preference, you can use this option, where you match user preference present in user attributes with the item attribute.
For example - Filter items (restaurants) where the Restaurant Ratings is more than 4, the Restaurant category is the same as the user's preference (preference category) and the restaurant location is within 5000 meters of the user's location (last_known_location).
Order Item Result
You can order item results by any of the numeric item attributes in ascending or descending order.
Order by Most recent option is available only for filters based on User Behavior, which sorts the item based on the recent action performed on those items.
Recommended Items
This is powered by the Sherpa AI engine that helps you build recommendations leveraging intelligent deep-learning algorithms from the no-code platform in just less than 24 hours. Select the relevant user actions you want to feed to AI recommendations and the user-personalized relevant recommended items are generated which you can use in any campaign. Please find more details here
View the created recommendations under the All Recommendations tab.