SFMC Tips #341 : Marketing Cloud Next: Data 360 Marketing Agent

Photo by Oliver Sjöström on Unsplash

In the September new feature release for Data 360, a new agent called Data 360 Marketing Agent was introduced to support segment creation.

• This is not a feature limited to Marketing Cloud Next. Since it is an agent for Data 360, it can also be used, for example, with segments for activation to Marketing Cloud Engagement.
• Since the agent type is Employee, please grant agent access permissions to the users and profiles that will use it.

Differences from Einstein Segment

Marketing Cloud Next / Data 360 has already had Einstein Segment, which uses generative AI to create segments from natural language.

The main functionality of the existing Einstein Segment is:

• Creating segments from natural language only

In contrast, Data 360 Marketing Agent can perform:

• Creating and modifying segments
• Checking segment populations
• Publishing segments
• Checking segment health
• Investigating why publishing failed
• Checking whether a specific customer is included in a segment
• Investigating why a specific customer is not included in a segment
• Checking segment billing size
• Checking overlap between multiple segments
• Creating a new segment from overlapping customers

In other words, it provides a mechanism that allows an Agent to perform management and investigation even after a segment has been created.

In Data 360 Marketing Agent, processing is broadly divided into the following three areas by subagent.

① Segment Builder Subagent

• A function for creating new segments and adding or modifying conditions in existing segments

② Segment Management Subagent

• A function for checking and managing the population, publishing operations, health, billing size, membership, and other aspects of existing segments

③ Segment Overlap Subagent

• A function for comparing multiple existing segments and checking overlapping customers

Note that Einstein Segment itself is still listed in Salesforce’s official documentation, so at this point, Data 360 Marketing Agent should not be considered a complete replacement for Einstein Segment.

Segment Builder

First is Segment Builder, which creates and modifies segments using natural language.

With Data 360 Marketing Agent, for example, you can make a request in natural language such as:

• “Create a segment of customers who live in Tokyo and purchased a product within the past 30 days.”

Internally, conditions are extracted from the entered natural language, and segment conditions are generated through the following flow:

Natural Language
↓
Conditions
↓
DMO / Attribute
↓
Relationship
↓
Segment Filter
↓
Segment

The conditions for the attributes to be used are determined in detail through natural language, as shown below.

① Supports AND / OR / EXCEPT

It supports not only simple conditions but also segments that combine multiple conditions.

For example:

• “Customers who live in Tokyo or Kanagawa and made a purchase within the past 30 days. However, exclude customers who have canceled their membership.”

Data 360 Marketing Agent extracts an AND / OR / EXCEPT logic tree from natural language in addition to the individual conditions.

Also, if the grouping of conditions is ambiguous, the Agent is designed to ask the user for clarification.

Note: According to the official Einstein Segment documentation, the “OR” and “Exclude” operators cannot be used in prompts.

② Create Segments Across Related DMOs

In actual segments, conditions do not always consist only of data from Individual or Unified Individual.

For example, for a condition such as:

• “Customers who purchased a product within the past 30 days”

it is necessary to follow a Relationship from Individual to a related DMO such as Order.

Data 360 Marketing Agent also includes processing to explore related DMOs and identify the appropriate path.

Therefore, even if the user does not fully understand all of the Relationships, it is designed so that the Agent can build a segment while searching for the relevant DMO, Attribute, and Relationship.

③ Modify Existing Segments Using Natural Language

Data 360 Marketing Agent can not only create new segments but also modify the conditions of existing segments.

For example:

• “Add a condition to this segment for customers who clicked an email within the past 90 days.”

In this case, instead of rebuilding the entire segment from scratch, it generates the conditions to be added or modified and merges them with the existing conditions.

Note: If there are unsaved conditions currently being edited on the Segment Builder canvas, the design also merges the new conditions with them rather than replacing them.

④ Supports Nested Segments

It also supports nested segments, where an existing segment is used as a condition for another segment.

For example:

• “Customers who are included in the VIP Customers segment and have not made a purchase within the past 30 days.”

Data 360 Marketing Agent can treat an existing segment as one condition and generate a segment by combining it with regular Attribute conditions.

In other words, you can reuse segments that have already been created and use natural language to create a segment that combines additional conditions.

Segment Management

This is where Data 360 Marketing Agent becomes particularly interesting.

Creating a segment from natural language can already be done with Einstein Segment, but Data 360 Marketing Agent allows you to manage and investigate created segments through the Agent as well.

Let’s take a look at what it can do.

① Check the Segment Population

Data 360 Marketing Agent can check the Population included in a segment.

For example, you can ask:

• “How many people are in this segment?”

to check the total population of the segment.

Get Nobuyuki Watanabe @marketingcloudtips’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

You can also retrieve the following information as needed:

• Population Breakdown: Check the population for each Filter
• Preview: Preview the Profiles included in the segment

This feature can be used to check whether the segment has the expected size and target audience before publishing it.

② Publish a Segment

Segments can also be published from Data 360 Marketing Agent.

The following publish types can be specified:

• Immediate: Publish immediately
• Scheduled: Publish according to a specified schedule
• Manual: Publish manually

You can also check the execution status of the publish job, including:

• RUNNING: In progress
• COMPLETED: Completed
• FAILED: Failed

In other words, operations such as:

• “Publish this segment.”

and checking the subsequent execution status can all be performed within a conversation with Data 360 Marketing Agent.

③ Diagnose Segment Health

One feature that I personally found interesting is segment health diagnosis.

Data 360 Marketing Agent includes an Observability feature for analyzing segment execution status.

In addition to the status of segment jobs, it is designed to detect issues such as the following and return the “failure reason” and “recommended action.”

• Data Explosion: A state in which following Relationships significantly increases the number of records to be processed, resulting in a high segment processing load
• Skew Join: A state in which the data used for a Join is extremely concentrated on specific values, placing a heavy load on some Join processing

In other words, you may be able to ask the Agent:

• “Why did this segment fail?”

and investigate the cause.

When a segment becomes complex, simply investigating why publishing failed can take a considerable amount of time. If the Agent can assist with this, it seems that it could be quite useful for segment operations.

④ Investigate “Why Isn’t This Customer in the Segment?”

This feature allows you to check whether a specific customer is included in a segment and, if not, investigate the reason.

When operating segments, there are cases where you need to investigate questions such as:

• “This customer should meet the conditions, so why aren’t they in the segment?”

With Data 360 Marketing Agent, you can perform checks such as the following:

• Check segment membership: Check whether a specific Profile is included in the specified segment
• Investigate why a Profile is not included in a segment: If a Profile is not included in the segment, identify which filter conditions it does not meet

For example, you could ask:

• “Is John Smith in the VIP Customers segment?”

and, if he is not included, continue by asking:

• “Why isn’t he in it?”

Furthermore, rather than simply responding that “the conditions are not met,” the system is designed to identify the filter that excludes the Profile from the segment and highlight the filter causing the issue in the UI.

In other words, investigations that previously required a person to check the segment conditions and Profile data one by one may be delegated to Data 360 Marketing Agent.

This means that it can be used not only as a Segment Builder but also as an Agent that supports segment troubleshooting.

⑤ Check Segment Billing Size

Data 360 Marketing Agent also includes a feature for checking in advance the scale of records that may be processed when a segment is executed.

With Segment Billing Estimate, Data 360 Marketing Agent can check the scale of records to be processed in the following three categories:

• SMALL: Up to approximately 20 million records
• MEDIUM: Approximately 20 million to 200 million records
• LARGE: More than approximately 200 million records

What is shown here is not the final number of customers included in the segment, but the scale of the number of records processed when calculating the segment.

For example, even if the final segment population is not very large, if the conditions reference multiple DMOs or a large number of related records, many records may be processed during segment calculation.

Therefore, by asking Data 360 Marketing Agent:

• “What is the billing size of this segment?”

you can understand the approximate processing scale of the segment before publishing it.

This can be thought of as a feature that provides a guideline for checking the processing scale before execution when creating a complex segment.

Segment Overlap

Data 360 Marketing Agent also includes a Segment Overlap feature that compares multiple segments and checks how much the customers included in those segments overlap.

For example, Data 360 Marketing Agent can perform operations such as:

• Check Segment Overlap: Compare two or more segments and check how much their populations overlap
• Create a new segment from an Overlap: Use customers included in both of two segments to create a nested segment

For example, if you ask:

• “How much do VIP Customers and High Engagement Customers overlap?”

you can check the overlap between the populations included in each segment.

Furthermore, by requesting:

• “Create a new segment with customers who are included in both.”

you can also create a new segment using the overlapping portion of the two segments.

For example, by examining the overlap between “customers with high purchase amounts” and “customers with high email engagement,” you could identify customers who make a high contribution to sales and also actively respond to marketing initiatives.

Rather than simply creating segments individually, this can be thought of as a feature that allows you to analyze the relationships between existing segments and use the results to create new audiences.

Additional Note: Not All Segment Types Are Supported

Data 360 Marketing Agent cannot create all segment types available in Data 360 from natural language.

Based on what I confirmed this time, Conversational Segment Builder can create standard publish segments and rapid publish segments, while segment types such as the following are currently not supported for creation:

• Waterfall
• Realtime
• Dynamic
• Look-a-like

Therefore, Data 360 Marketing Agent does not replace all of Data 360’s segmentation functionality.

It is easier to understand it as a mechanism that supports natural-language segment creation and modification, as well as subsequent management and investigation, for the segment types that it supports.

What did you think?

When I first looked at Data 360 Marketing Agent, I thought:

• “Is this just a new Agent that can create segments using natural language?”

However, it turned out to be much more than that.

Data 360 Marketing Agent is designed to cover a fairly broad range of the segment lifecycle, including not only segment creation but also checking populations, publishing, health diagnosis, membership investigation, billing size, and overlap analysis.

Therefore, rather than simply being a segment creation Agent, it is easier to think of it as an Agent that supports everything from segment creation to operation and investigation.

Please give it a try.

That’s all for this article.

Stay tuned for more Salesforce Marketing Cloud tips! 😎

Nobuyuki Watanabe

https://www.linkedin.com/in/nobuyuki-watanabe/ (+Follow me)