Overview

Segmentation provides a structured and organized approach to identifying and querying attributes within an object. This enables effective analysis and decision-making based on the characteristics of the object. For more information, refer to Support for Object Data Type.

• On the Create segment page of the MoEngage dashboard, the user segmentation capability of an object data type is displayed under the User property and User behaviour tabs of Select audience section.
• Segmentation supports two levels of nesting for the object data type. Above level two nesting, the attributes are shown in the stringified format.
  
• In case of an array of object data type attributes, segmentation will provide the With operator and the And operator for advanced querying.

Create Segmentation Queries with Objects

The attributes of an object can be of any primitive data type. This section provides few examples of writing segmentation queries for user attribute and event attribute scenarios using the following data types:

Query String Attributes

User Attribute

If you want to segment users who have pet dogs, you can use the string data type Animal.pets and the operator is to fetch users who have a dog as their pet.

Event Attribute

If you want to segment users who have Adopted a pet at least once in the last 3 days with attributes such as Membership (String data) as Gold and users having a pet as Cat (Object data).

Description:
Here, Membership is an event attribute with a string data type and Pets.species is another event attribute with a string Object data type. The And operator ensures that the users fetched from the database have fulfilled both conditions.

Query Empty String Values Using Segmentation

If you want to segment users who have an empty string value for a specific attribute. The stored value is "" (The key is set, but the value is null).

Description:
Here, ListOfPets is an object with Name as a user attribute inside it. Thus in the query, the ListOfPets.Name attribute is used with the is empty filter.

Query Numeric Attributes

User Attribute

If you want to segment users who have pets younger than 6 years.

Description:
Here, Pets.age is the user attribute with a numeric data type and less than is the relational operator to filter users with pets aged less than 6.

Event Attribute

If you want to segment users who have purchased a t-shirt that costs $500.

Description:
Here, the User behaviour drop-down displays the Object attribute product.price and the value as $500. The segmentation query has the following event attributes:

• Product.Colour as the attribute with string data type
• Product.Price as the Object attribute with numeric data type
• Product.Currency as the Object attribute with string data type

[
{ product: “t-shirt”
price: “$500”
colour: “red”
currency: “USD”
}
 { product: “trouser”
price: “$800”
colour: “white”
currency: “EUR”
}
{ product: “t-shirt”
price: “$300”
colour: “blue”
currency: “EUR”
}
]    

Query Date Time Attributes

Event Attribute:

If you want to segment users who have purchased a red t-shirt whose price is $500 and whose purchase date is before 30 May 2024,

Description:
Here, the segmentation query has the following Object event attributes:

• Colour is an event attribute with string data type Red.
• Product.Price is an event attribute with string data type.
• Product.Purchase-date is an event attribute Object of date data type

[
{ product: “t-shirt”
price: “$500”
colour: “blue”
purchase-date: “23-12-2023”
}
 { product: “trouser”
price: “$800”
colour: “white”
purchase-date: “23-12-2023”
}
{ product: “t-shirt”
price: “$300”
colour: “red”
purchase-date: “23-12-2023”
}
]

Query Array Attributes

Event Attribute

If you want to segment users whose dogs were vaccinated at least 1 time in the last 60 days, breed as Lab, age under 5 years, and bone as their favourite treat.

      Pets object array: 
[ 
{ species : “Dog”,
vaccinated: “Yes”     
Age : 2,    
treats : [‘bone’, ‘candy’, ‘pedigree’]  
},  
{ species : “Cat”,
vaccinated: “Yes”    
Age : 2,    
treats : [ ‘candy’, fish]  
}, 
{ species : “Dog”, 
vaccinated: “No”    
Age : 4,   
treats : [‘bone’, ‘pedigree’]  
},
]

    

Query Nested Objects

Event Attribute

If you want to segment users who have purchased a book whose price is 300 USD.

[
{
  "item-id" : 123,
  "item-type" : "books",
  "item-cost" : {
    "amount" : 100,
    "currency" : "USD"
  },
{
  "item-id" : 456,
  "item-type" : "fruits",
  "item-cost" : {
    "amount" : 200,
    "currency" : "USD"
  },
{
  "item-id" : 678,
  "item-type" : "books",
  "item-cost" : {
    "amount" : 140,
    "currency" : "USD"
  }
]

The Query Results section in the Create segment page displays the history of all queries executed on the dashboard. For each query, you can find the details such as Query Time, Description, Source, User Count, and Reachable Users.

Functionality of the “With” Operator

You can use both “With” and AND/OR operators to create queries using the Object data type. The AND/OR operator applies the attribute filters across the objects, whereas the With operator applies the attribute filters within an object.

Using the With Operator

Consider the following pet object data type as an example of using the With operator to achieve some uses cases below.

[
{ Breed: “Lab”
Age: “3”
Treat: “Bone”
},
{ Breed: “Lab”
Age: “5”
Treat: “Fish”
},
{ Breed: “Bulldog”
Age: “3”
Treat: “Pedigree”
}
]   

Example 1

If you want to segment users whose dogs were vaccinated 3 times in the last 60 days, whose dogs should be of the breed Lab, with their dogs' age less than 5 years old, and with the dog's favourite treat being Bone, using the With operator:

To achieve this use case, break down the query as listed below:

• Find the users whose pets are dogs of breed Lab
• Amongst the Lab breed, find those dogs whose age is < 5 years
• Amongst the Lab breed of < 5 years, find those dogs whose favourite treat is Bone

Here, the with operator will first search for a parent attribute (in the example above - it is pets.age) and then apply the filter to its child attributes (in this case - pets.treat).

This query will return the user mentioned above because in the array of objects, there is a Lab whose age is < 5 with their favourite treat as Bone.

Example 2

If you want to segment users whose dogs were vaccinated 3 times in the last 60 days, whose dogs are of the breed Lab, with their dog age less than 5 years old, with whose dog’s favourite treat is Fish.

This query will not return any user mentioned above because though there is a Lab with age < 5 years, their favourite treat is not Fish. Simultaneously, there is a Lab breed whose favourite treat is Fish, but its age is not less than 5. Thus, this user will not be returned during segmentation.

Using the And Operator

Consider the following pets object data type as an example of using the And operator to achieve some of the use cases below

{ Breed: “Lab”
Age: “3”
Treat: “Bone”
},
{ Breed: “Lab”
Age: “5”
Treat: “Fish”
},
{ Breed: “Bulldog”
Age: “3”
Treat: “Pedigree”
}
]   

Example 1

If you want to segment users whose dogs were vaccinated three times in the last 60 days, whose pets are dogs of breed Lab AND users with pets whose age is > 3 years AND users whose pets have their favourite treat as Fish.

This query will return those users who:

• Have pets that got vaccinated 3 times in the last 60 days
• Have pets that are dogs of breed Lab
• Have any pets whose age > 3 years
• Have any pets with their favourite treat as Fish

Thus, the given user will be returned during segmentation because somewhere in the entire array of objects, there is a Lab-breed pet, and some pets are older> than 3 years. Also, some pets have Fish as their favorite treat. Thus, all three conditions are satisfied independently, as required in the case of the AND condition.

Using the OR Operator

Consider the following pets object data type as an example of using the OR operator to achieve some of the use cases below 

{ Breed: “Lab”
Age: “3”
Treat: “Bone”
},
{ Breed: “Lab”
Age: “5”
Treat: “Fish”
},
{ Breed: “Bulldog”
Age: “3”
Treat: “Pedigree”
}
]   

Example 1

If you want to segment users whose dogs got vaccinated three times in the last 60 days, whose pets are dogs of breed Lab OR users with pets whose pets have age is > 3 years OR users whose pets have their favorite treat as Duck.

This query will return the user mentioned above because it fulfils the definition of the OR operator (which says that at least one of all the conditions must be fulfilled). In this case, only favorite treat condition = Duck is not fulfilled, but the other two conditions are fulfilled, which will return the user above during segmentation.

Key Factors for Object Data Type Ingestion

• The object data type is not supported in the File segment upload.
• User export through CSV does not support object user attributes.
• Aggregation of the array of Objects is not supported. However, aggregation of objects and nested objects is supported.
• Object-to-object attribute comparison is not supported in user attribute comparison. Object-to-primitive attribute comparison is still supported.
• The system automatically chooses the first data type and performs the query if multiple data types are ingested for an attribute.
• Redirection from the Create segment page to the campaigns or analytics page is not supported while creating segmentation queries. You must create a custom segment that can be used in all campaigns and analytics pages.
• Currently, if both an object and an array of object data type values are passed for an attribute, the MoEngage system will consider the first data type that is passed. It is not possible to query the second data type value. As a workaround, the customer has to use two separate attributes, set them as objects in data management, and pass the object and array of object values separately in those two attributes.