Advanced Analytics for SAP Marketing Cloud
In this article, you will learn about the advanced analytics terminologies, roles, and advanced analytics scenarios in SAP Marketing Cloud.
Table of Contents
Overview About Advanced Analytics
In the area of Intelligent Marketing, there are many terminologies such as Advanced Analytics, Predictive Analytics, Machine Learning or Artificial Intelligence. Sometimes these terms are used interchangeably, but in fact they are not synonymous. Let's review the terminologies that are used in this article.
image one: Advanced Analytics Terminology
Within Advanced Analytics, there are three techniques or disciplines:
- Predictive Analytics: Employs technologies such as statistical modeling or simulation
- Prescriptive Analytics: Includes optimization, heuristics and rule-based expert systems following business rules
- Artificial Intelligence: A subset of Advanced Analytics that includes machine learning, natural language processing, and cognitive advisors
SAP Marketing Cloud covers scenarios and features within the Prescriptive Analytics (Rule-based Scores/Score Builder) and Artificial Intelligence (Predictive Scores, Recommendation, Sentiment Analysis) areas.
Roles for Advanced Analytics
To break down the complexity of Advanced Analytics, let's look at the typical roles within Advanced Analytics projects or sub-projects.
The business users are responsible to utilize the results to target consumers provided by business analysts. Typical examples are:
- Use predictive "buying propensity score" in segmentation to target consumers which are most likely to buy a specific product
- Provide feedback on product recommendation model results to business analysts or data scientists
The business analysts use, adjust, and enrich analytical models based on standards typically provided by the data scientists. Some examples are:
- Fine-tune offer recommendation models to provide better results based on business user feedback
- Select predictors for predictive score based on model quality
The data scientists or data analysts perform complex analytics and develop standards mainly for business analysts.
- Evaluate best algorithms for use in recommendation models (data scientist)
- Support business analysts in fine-tuning of recommendation models (data scientist)
- Build predictive scenarios including own data sets with pre-defined algorithms (data analyst)
image two: Advanced Analytics Roles
Advanced Analytics in SAP Marketing Cloud
The following table explains which scenarios in SAP Marketing Cloud can be mapped to which technique and role.
|Scenario||Technique||Involvement of Business Users||Involvement of Business Analysts||Involvement of Data Scientists or
|Rule-Based Scores||Prescriptive Analytics (Rule-Based Expert Systems)||Utilizing the scores for targeting consumers||Building and fine-tuning simple or complex rule-based scores||-|
|Predictive Scores||Artificial Intelligence (Machine Learning)||Utilizing the scores for targeting consumers||Adjusting and fine-tuning standard predictive scores||Developing custom predictive scores and predictive scenarios|
|Offer and Product Recommendation||
Artificial Intelligence (Machine Learning)
|Previewing recommendations and providing feedback on recommendation quality||Adjusting and fine-tuning standard recommendation algorithms||Developing custom recommendation algorithms|
Artificial Intelligence (Natural Language Processing)
|Utilizing the sentiment data for targeting consumers||-||-|
The next chapters cover these scenarios in more detail.
Scores are essentially key figures or aggregated values that help to characterize or classify the interaction contact for different aspects. See the following example for a score on account level - the account engagement score.
image three: Scores in SAP Marketing Cloud
In general, there are two types of scores in SAP Marketing Cloud - predictive (propensity-based) scores and rule-based (heuristic) scores.
|Rule-Based Score||Predictive Score|
|Concept||Based on rules and conditions||Based on predictive modeling|
|Learning||Best Practice: Rule sets are defined based on company policies, experience, and best practice. A rule set can contain several rules. Each rule consists of multiple conditions.||Data-Driven: The predictive model is trained on historical data to detect patterns in customer behavior.|
|Score Value Calculation||Calculated as aggregation of the outcomes from all valid rules. A rule is valid if all conditions are met.||Calculated from trained predictive model|
|Target Role||Business Analyst||
Business Analyst or Data Scientist
|App||Score Builder||Predictive Studio|
As explained above, rule-based scores are explicitly defined by best practice. Some real-life examples for rule-based scores are:
- E-mail Affinity (standard)
- This score represents the response affinity to the e-mail channel. The score is calculated by looking at the e-mail open rate and if an e-mail opt-in exists among others.
- Best Push Notification Sending Time (standard)
- This score helps to send push notifications at the right time. The score represents the "best sending time" by looking at the peak times when push notifications are most viewed by the contact.
- Lead Quality Score (custom)
- A lead quality score can tell you how well you know your contacts based on data completeness.
- RFE Score (custom )
- A RFE score can help you to measure loyalty or brand value by looking at the three dimensions recency, frequency, and engagement.
An article about RFE scoring for SAP Marketing Cloud is available here. This article explains in detail the use cases for an RFE score, and how such a score can be implemented and utilized in SAP Marketing Cloud.
All rule-based scores are built and provided to the business users through the Score Builder app. This app is explained in the following video:
In a nutshell, predictive scores are trained through predictive models on historical data. The idea is to detect underlying patterns in the customer's behavior. The predictive models are then used to score interaction contacts. Some real-life examples are:
- Consumer Buying Propensity (standard)
- This score indicates how likely it is that a consumer is going to buy a specific product. The score is calculated based on their interactions within a given time period among other attributes.
- Insurance Churn Propensity(custom)
- Such a score can indicate which customers tend to cancel their insurance policy. This score is calculated based on customer master, product and interaction data within a given time period.
An article about custom predictive scores for SAP Marketing Cloud is available here. This article explains in detail the use cases for custom predictive scores, and how such scores can be implemented and utilized in SAP Marketing Cloud.
All predictive scores are built through the Predictive Studio app. The following video will help you to understand how to use this app.
Recommendations allows you to provide consumers with relevant recommendations for offers and products across multiple channels. Business analysts can work on recommendation models based on standard recommendation algorithms or on custom algorithms provided by data analysts or data scientists. Recommendations are always provided for a certain context. For example:
- For specific consumers (based on preferences or order history)
- For certain products (such as the products currently viewed or in the shopping cart)
image four: Recommendations (in bottom section)
- Recommend products often bought together with the products in the shopping cart (standard)
- Recommend more expensive products from the same product category (custom)
- Recommend eligible offers based on leading products (standard)
- Re-rank offers based on expected margin (offer)
An article about custom recommendation algorithms for SAP Marketing Cloud is available here. This article explains in detail the use cases for custom recommendation algorithms, and how such algorithms can be implemented and utilized in SAP Marketing Cloud.
The following video shows you how you can integrate SAP Marketing Cloud with SAP Commerce Cloud to provide product and offers recommendations for a personalized shopping experience. Product recommendations are discussed after 3:12 and offer recommendations are discussed after 6:41:
Sentiment Engagements helps to automatically analyze interactions containing textual information such as social posts or service tickets. If the interactions are marked as relevant for text analysis, they get analyzed automatically by SAP HANA Text Analysis asynchronously. The resulting sentiments (positive, neutral, negative) and entity types are extracted and assigned to the interaction.
Please notice that currently you can't influence the underlying text analysis logic, for example, by defining customer-specific tags which might represent product categories or interests.
This article introduced you to advanced analytics for SAP Marketing Cloud. Now, you should have an overview about the terminologies, roles, and advanced analytics scenarios in SAP Marketing Cloud.
The articles below will help you dive deeper on advanced analytics topics:
- Measure Brand Love and Customer Value in SAP Marketing Cloud Using a RFE-Based Score
- How to Implement Custom Predictive Scores for SAP Marketing Cloud
- How to Implement Custom Algorithms for Product Recommendations in SAP Marketing Cloud
If you are interested in learning more about advanced analytics for SAP Marketing Cloud, we also offer a Advanced Analytics Guidance service.