Client, staff member, and partner expectations have developed an overwhelming burden on business directors and heads to innovate by building new modes of value and communication. By utilizing predictive analytics, businesses can anticipate results, recognize prime opportunities, reveal unseen risks, predict the future, and perform in the here and now.
Profitable businesses appreciate the power of applying predictive analytics across business systems, functions, and line-of-business offerings to keep up the competitive edge. SAP Marketing Cloud offers you the innate benefit of predictive analytics with functions such as segmentation, campaigns, and so on to bring the above mentioned goals to successful completion.
The major advantages of using Predictive Analytics in the SAP Marketing Cloud include:
- Embed and extend predictive results into applications
- Optimize processes and performance
- Detect patterns to initiate campaigns
- Aggregate and correlate information
- Catch the best trends for better ROIs
- Faster and accurate outcomes
- Scale and manage the end-to-end lifecycles
Why we need predictions in today’s advanced marketing automation
Though today’s marketing automation tools are far more advanced with their tracking and real-time capabilities as compared to the traditional marketing tools, studies prove that the addition of predictive analytics in marketing applications leads to the best/highest ROIs (as much as 250% as compared to regular campaigns).
With the addition of the Predictive Analytics Library in the SAP Marketing Cloud system, you can now apply statistical means to discover designs and flow in historical data and employ the information to anticipate customer actions.
Quality master data for effective predictions
For the best predictions, provide as much detail on the contacts as possible; this helps the models to train better so that they can predict better. Any of the below information can help separate the groups:
- Marital status
- Sales current year
- Sales previous year
- Change of sales between years
- Latest contact details
- Number of contacts/visits
- Payment history
- Outstanding payment
Predictive Analytics in SAP Marketing Cloud
Below are the few areas where you can use predictive analytics in SAP Marketing Cloud:
- Segmentation/Target Groups
- Email Campaigns
Architecture & Components
SAP Marketing Cloud’s Predictive Analytics are built on APL (Automated Predictive Library), which is now part of SAP HANA system. The below diagram shows the detailed architecture and the associated components in the SAP Marketing Cloud.
Segmentation with Prediction KPIs
SAP Marketing Cloud’s market-leading segmentation capability allows users to segment their target markets with predictive Key Performance Indicators (KPIs) such as:
- Profile scores/Top-Ranked Customers
- Buying propensity
- Cancellation propensity
- Selected customers
- Coverage of potential buyers
- Predicted potential buyers
The predictive KPIs are established on predictive models that allow keeping count of customers. The goal is to communicate with a portion of customers in the best dimensions of the score, permitting the most advantageous number of potential clients, contrary to addressing a chance choice of customers.
The effects of predictive segmentation models can unequivocally be applied by campaign managers to perfect the magnitude of the target group and spontaneously communicate with every contact utilizing the most excellent channels with a customized message by - maybe – something like an e-mail, letter, or phone.
An example of a segmentation model using predictive KPI Buying Propensity is shown below:
Predictive Analytics in Email Campaigns
Predictive analytics takes advantage of existent email campaigns and connects real-time behavior identification with historical customer information to develop automated, customized email campaigns. This helps the email campaigns to a great extent and –allows teams to send emails that convert more prospects into customers, and send emails that engage individual customers and reengage customers who haven’t purchased from you in a while.
Ultimately, these tactics help marketers save time and also gain more accurate results by sending the coupons and offers to their best prospects.
The below screenshot shows how Email Affinity scores can be used in SAP Marketing Cloud Campaign Automation:
All Hybris Marketing applications use predictive analytics that’s built and trained on profile scores.
The profile score is a formula to calculate a value for a given question; for example, how loyal is a customer in a standardized way? The value denotes the total score calculated for the given contact. Examples of these scores include loyalty scores, buying propensity, email sending time, and so on.
The scores are to be calculated based on various inputs such as Master Data, Behavioral Data, Historical Data, and a combination of all these inserted in Rule Sets and Rules.
There are two types of scores:
- Rule-Based Scores: Rule-Based scores are based on best practices. This means that the rule sets are defined based on company policy, experience, and best practices. A rule set has various rules and every rule includes various conditions. The Score Builder application is used to build rule-based scores.
- Propensity Scores: Propensity scores are based on predictive modeling and are data-driven. This means that the predictive model is trained on historical data to detect patterns in customer behavior. The Predictive Studio application is used to build propensity scores.
Score Builder application (to build rule-based scores)
We can use these scores as attributes in segmentation applications:
Predictive Studio (to build Rule-Based scores on the SAP Marketing Cloud)
Predictive Studio is used to create, train, and maintain predictive models for the calculation of Key Performance Indicators (KPI). These predictive KPIs can be used in multiple applications; for example, they can be used in segmentation or email campaigns.
The below list of tasks can be achieved with the help of Predictive Studio on the SAP Marketing Cloud:
- Filtering According to Ownership
- Filtering According to Model Status
- Searching for a Model
- Creating a Model
- Processing an Existing Model
- Deleting Predictive Models
- Defining and Publishing a Predictive Model
- Fitting a Predictive Model
Steps/Best practices to create a predictive model
- Define a predictive model in SAP Marketing
- Define an applicable scope; ex: Country or Region. The available attributes for an applicable scope are defined in Customizing
- Define the context of the model, select a predefined predictive scenario; for instance, Buying Propensity, which includes the data source, the use case, and the applicable algorithms (implementation methods) for the predictive calculation
- Select a Campaign for Success Measurement
- Create a Model Fit
- Set the status to “In Preparation”
- Enhance the model (Edit an existing model) with model fits and scope
- Create a model fits with different implementation methods, different sets of predictors, and cross validation
- Start model training to train the model. Check for training results such as Predictive Power, Prediction Confidence, Initial No. of Predictors, No. of Selected Predictors, Reference Products as Predictors, and No. of Kept Predictors
- Chose the best fit to publish the model. Use Lorenz Curves or The Box Plot to check for the best fit
- Publish the model
- To restrict the validity of the model, define a set of countries and regions as scope for which the model was valid
- To withdraw a published model, set the status to “Complete”
- Describe the model properly by using the Key Information and Notes facet to add the description and personal notes about the model
Below are some screenshots for you to see what Predictive Studio looks like in the Marketing Cloud application:
Predictive Scenario Settings in the backend (in SAP Marketing On-Premise)
Prior to creating a predictive model in the front end, the predictive scheme has to be elucidated in IMG customizing. A predictive scenario describes the framework of the predictive model, which contains the data origin, the use case, and the appropriate algorithms (implementation means) and others.
All IMG activities related to predictive scenarios can be defined in:
IMG > SAP Hybris Marketing > Data Management > Predictive Scenarios
The primary group of predictors can be described in Customizing for SAP Marketing thus: Predictive Scenarios > Define Predictive Data Source
The front-end UI settings can be defined thus: Predictive Scenarios > Define Settings for Display of Predictive Scenarios on UI
The predictive data source is called to by the script-based estimation view sap.hana-app.cuan.cpred.demo.datafoundation/CA_CPRED_BANKING_DS_PAR. It provides the business data for the training set, which contains data that describes the actual customer behavior. It is then used to detect patterns and trends for future prediction. The cross-authorization and preparation of predictive models are established on a group of SQL script procedures, which are applied to call PAL functions to be employed for predictive calculation.
Again, as per SAP’s future road map, this is just a beginning and we will see more focus and more applications coming soon that will be closely associated with predictive analytics.
My take on predictive analysis in marketing and more
It’s only going to get more crowded with predictive analytics out there. If you are a larger business, know that smaller businesses can and will also start to take advantage of predictive analytics too. If you are a smaller business, know that you can compete with some of your larger competitors with predictive analytics. However, the singular method by which you will attain an amazing outcome is by developing an appealing customer experience that could only come from you – not by copying someone else.
Are you using predictive analysis? If so, how’s it going? If not, why? Let me know by writing to me at firstname.lastname@example.org.