Android integration:: Mobile Engage - Predict Product Recommendations – Emarsys

Contents

What are Predict Product Recommendations
Product Recommendations AI Logics
Recommendation Filters
Set Limit
Result Listener
Click Tracking

Resources

Sample App

What are Predict Product Recommendations

Predict is a recommendation engine which analyses the behaviour data collected from your store by our JavaScript API (web) and Emarsys SDK (mobile) and uses this to deliver personalised recommendations to all your customers across email, mobile and web.

The Emarsys SDK supports this feature by providing the behavioural tracking in the standard SDK installation guide and the ability to retrieve/display the AI recommended products. The details on how to pull the product recommendations is documented below.

Product Recommendations AI Logics

The Emarsys SDK facilitates the request of product recommendations based on different recommendation logics used within the call. Below is confirmation of the supported logics available.

recommendProducts is also going to track the value attached to the logic on the backend, so no additional tracking needed when using recommendations!

Suggested App Screen	Logic	Description
Home	Home	Personalized recommendations for product categories and products within these categories.
Product	Related	Related/similar product recommendations.
Product	Also_Bought	Complementary product recommendations for the currently viewed product.
Cart	Cart	Complementary product recommendations matching the cart contents.
Category	Category	Personalized recommendations in a given category.
Search Results	Search	Personalized recommendations matching the search term entered.
Home	Personal	Personalized recommendations to match current recent behavioural tracking of the contact.
Category	Popular	A simple list of top-selling products in a given category.

It is possible to pass the values to the chosen recommendation logic. In the event that the values are not supplied in the call, the Emarsys SDK uses the last tracked values.

Recommendation Filters

This is an optional parameter of the recommendProducts method.

You can filter product recommendations with the SDK by building RecommendationFilters. There are two types of filters: Exclude or Include.

In every case there are four types of comparators you can use to compare your chosen field to expectationValue:

isValue- checking if the field is matching the value
inValues- any of the values has a match with the field
hasValue- One of the field values is equal to expectation value(applicable only to fields containing multiple values)
overlapsValues- One or more of the field values are found in expectation values(applicable only to fields containing multiple values)

For further information please check the predict documentation.

Set Limit

This is an optional parameter of the recommendProducts method.

You can limit the number of recommended products received by defining a limit. This is an optional parameter, by default its value is 5.

Result Listener

This is a required parameter of the recommend Products method.

The SDK is going to retrieve recommended products via the resultListener

Java

RecommendationFilter filter = RecommendationFilter.exclude("category").isValue("SHIRT");
List<RecommendationFilter> filters = new ArrayList();
filters.add(filter);
Emarsys.getPredict().recommendProducts(RecommendationLogic.category(), filters, new ResultListener<Try<List<Product>>>() {
    @Override
    public void onResult(@NonNull Try<List<Product>> result) {
        if (result.getResult() != null) {
            List<Product> recommendedProducts = result.getResult();
        }
        
        if (result.getErrorCause() != null) {
            Throwable cause = result.getErrorCause();
            Log.e(TAG, "Error happened: " + cause.getMessage());
        }
    }
});

Kotlin

val filter = RecommendationFilter.exclude("category").isValue("SHIRT")
Emarsys.predict.recommendProducts(RecommendationLogic.category(), listOf(filter)){
    it.result?.let { result ->
        val recommendedProducts = result.result
    }
    it.errorCause?.let { cause ->
        Log.e(TAG, "Error happened: ${cause.message}")
     }
}

Click Tracking

The Emarsys SDK does not automatically track recommendationClicks, therefore it is required to call manually trackRecommendationClick when an interaction happens with any of the recommended products.

Java

Emarsys.getPredict().trackRecommendationClick(Product clickedProduct);

Kotlin

Emarsys.predict.trackRecommendationClick(clickedProduct: Product)

Variants

Variants are used by the HOME and PERSONAL logic types. By adding a list of Strings used as suffixes to logic names, recommendations are grouped by the variants provided.