Configuration Guide | PUBLIC

2025-07-02

Configuring SAP Intelligent Product

Recommendation
© 2025 SAP SE or an SAP affiliate company. All rights reserved.

THE BEST RUN

Content

1 Introduction to SAP Intelligent Product Recommendation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Preparing Data for Ingestion and Transformation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1 Creating Product Categories and Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Assigning Products to Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8
2.3 Creating Product Selection Profiles for Product Categories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
2.4 Synchronizing Descriptions of Attributes and Configurable Products. . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 Informing Machine Learning Specialist about IPSP Activation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3 Preparing Data for Machine Learning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.1 Creating Raw Datasets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
3.2 Importing Data to Raw Dataset. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Creating Engineered Dataset. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.4 Generating Document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18
3.5 Downloading Document to Local System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.6 Uploading Document. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
Data Extraction from SAP ECC System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Lead Time and Price Prediction Calculation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.7 Importing Document. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.8 Creating Material Variant Datasets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.9 Generating Document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.10 Uploading Material Variant Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.11 Importing Document. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
3.12 Viewing Runtime Dataset. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25

4 Preparing Machine Learning Profiles for Recommendations. . . . . . . . . . . . . . . . . . . . . . . . . . . 26

4.1 Creating Machine Learning Profiles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.2 Creating Machine Learning Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3 Training Machine Learning Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
The Best Machine Learning Models for Recommendations. . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.4 Deploying Machine Learning Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
4.5 Publishing Machine Learning Profiles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.6 Informing Product Manager About Published ML Profile. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

5 Activating Product Selection Profiles (IPSP) for Product Recommendations. . . . . . . . . . . . . . 32

5.1 Product Selection Profile (IPSP) Deactivation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

6 Built-in Support for SAP Intelligent Product Recommendation. . . . . . . . . . . . . . . . . . . . . . . . .33

Configuring SAP Intelligent Product Recommendation

2 PUBLIC Content
1 Introduction to SAP Intelligent Product
Recommendation

SAP Intelligent Product Recommendation is a cloud application that uses machine learning and rules
to streamline the product selection and configuration process for complex configurable products. The
combination of trained machine learning models and rules recommend products and default configurations
based upon the customer's specific needs. You can embed SAP Intelligent Product Recommendation in
configure, price, quote (CPQ), or e-commerce business processes. SAP Intelligent Product Recommendation
supports single-level configurable products.

This application is built on SAP BTP with integration to SAP S/4HANA as important sources of sales
transactional data.

The benefits for manufacturers are:

• Improve margin, increased revenue by reducing time and effort to create quote, reducing training costs,
fewer lost orders.
• Increase revenue by getting new sales representatives selling sooner by reducing required level of training
• Reduce customer lead time, improve on-time delivery by reducing variability in ordering and allowing more
standardization of components.
• Increase image and value of IT function, get closer to the business, reduced dependence on few key
resources, support Digital Transformation initiative.
• Allow manufacturers to increase revenue by expanding/scaling into additional global markets that were
previously ignored because of experience prerequisites for sales team.

Configuring SAP Intelligent Product Recommendation

Introduction to SAP Intelligent Product Recommendation PUBLIC 3
SAP Intelligent Product Recommendation Configuration Flow

• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]

Configuring SAP Intelligent Product Recommendation

4 PUBLIC Introduction to SAP Intelligent Product Recommendation
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]
• [Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?
state=DRAFT [[Link]
7e21c4f1857e4ee99699a4e7a9c9702a/[Link]?state=DRAFT]

Configuring SAP Intelligent Product Recommendation

Introduction to SAP Intelligent Product Recommendation PUBLIC 5
Next task: Preparing Data for Ingestion and Transformation [page 7]

Configuring SAP Intelligent Product Recommendation

6 PUBLIC Introduction to SAP Intelligent Product Recommendation
2 Preparing Data for Ingestion and
Transformation

Follow the steps to prepare your data by defining your requirements, and removing any unnecessary
information.

1. Creating Product Categories and Solutions [page 7]

Product categories and solutions are the building blocks in SAP Intelligent Product Recommendation.
2. Assigning Products to Solutions [page 8]
The assignment of products to solutions helps in determining the best technical configuration for a
product recommendation.
3. Creating Product Selection Profiles for Product Categories [page 9]
You can define the relationship between customer needs or requirements and the technical
specifications of products using the Intelligent Product Selection Profile (IPSP).
4. Synchronizing Descriptions of Attributes and Configurable Products [page 12]
Synchronize descriptions of attributes and configurable products that are associated with the IPSPs
with the source systems so that these descriptions and names can be displayed in the runtime
application in different languages.
5. Informing Machine Learning Specialist about IPSP Activation [page 13]
Let your machine learning specialist know about your IPSP that needs to be activated.

Previous: Introduction to SAP Intelligent Product Recommendation [page 3]

Next task: Preparing Data for Machine Learning [page 14]

2.1 Creating Product Categories and Solutions

Product categories and solutions are the building blocks in SAP Intelligent Product Recommendation.

Context

Product categories, solutions, and the configurable products assigned to these solutions help in determining
the needs attributes that are used by the runtime application. A guided selling expert uses the runtime
application to check and decide on the best product that matches their business requirements. Product
categories contain the functional needs of a desired product, and solutions contain the technical configuration
for the same.

 Example

Take the example of a sales quote for an industrial pump that is generated by SAP CPQ. The purchaser
knows their functional requirements, such as, characteristics of the pumped medium, the liquid's name,

Configuring SAP Intelligent Product Recommendation

Preparing Data for Ingestion and Transformation PUBLIC 7
 concentration, density, and viscosity, as well as duty point characteristics, like the desired flow rate, head
pressure, operating temperature, etc.

SAP Intelligent Product Recommendation decides the ideal technical setup for the pump based on the
requirements, like the housing specifications, mechanical sealing, and motor.

Procedure

1. Launch the Manage Product Categories application.

2. Choose Create.
3. In the flexible column that opens, enter a product category.
4. (Optional) Choose Create in the Descriptions section to maintain a product category description in the
required language.

 Note

The application displays the product category description in the header section in the logon language.

5. Choose Create in the Solutions section to create a solution for the selected product category.
6. In the last column that opens, enter a solution name.
7. Choose Apply.

Task overview: Preparing Data for Ingestion and Transformation [page 7]

Next: Assigning Products to Solutions [page 8]

2.2 Assigning Products to Solutions

The assignment of products to solutions helps in determining the best technical configuration for a product
recommendation.

1. Launch the Manage Product Categories application.

2. Select the required product category, and choose  navigation arrow right to open the details page.
3. In the details page, choose Edit.
4. In the Solutions section, select the required solution to assign the product.
5. In the last panel, choose Create in the Product Assignment section.
6. In the new field that is created, choose  value help to open the Product dialog box.
7. In the Product dialog box, select the required product from the list.
8. In the last panel, choose Apply.
9. In the details page, choose Save.

Configuring SAP Intelligent Product Recommendation

8 PUBLIC Preparing Data for Ingestion and Transformation
  Note

You can assign the same product to multiple Product Categories to support broader recommendation
scenarios. However, within a single Product Category, a product can only be assigned to one Solution.
Assigning the same product to multiple Solutions within the same category will result in a validation error.

Parent topic: Preparing Data for Ingestion and Transformation [page 7]

Previous task: Creating Product Categories and Solutions [page 7]

Next task: Creating Product Selection Profiles for Product Categories [page 9]

2.3 Creating Product Selection Profiles for Product

Categories

You can define the relationship between customer needs or requirements and the technical specifications of
products using the Intelligent Product Selection Profile (IPSP).

Context

You can define the needs attributes and technical configuration using the classification system characteristics
that are fetched from SAP S/4HANA. The products that are to be configured based on a specific set of
demands must be related to the attributes of the needs.

 Note

• Ensure that the description of an IPSP doesn't contain these characters: equal sign (=), plus sign (+), at
sign (@), tab (0x09) and carriage return (0x0D).
• If you want to edit an IPSP that is active, choosing Edit allows you to edit only the fields in the
Commercial Attributes Settings for Runtime Application section.
• You can add a sequence number to an attribute to define its order of display in the runtime application.

Procedure

1. Launch the Manage Intelligent Product Selection Profiles application.

2. Choose Create.
3. In the flexible column that opens, enter an IPSP name.
4. Choose  value help to select a product category in the Product Category field.

Configuring SAP Intelligent Product Recommendation

Preparing Data for Ingestion and Transformation PUBLIC 9
5. In the Needs Attributes section, choose Create to build a product selection profile from scratch.

The system creates an entry for you to assign the required attributes.

 Note

You can create multiple attributes for a product selection profile.

6. In the new entry that is created, perform the following steps.

a. Choose  value help in the first field to select a characteristic from the Characteristic dialog that gets
assigned as an attribute to the product selection profile.
b. In the next field, choose  navigation-down-arrow to select an attribute type for the assigned attribute.
c. In the third field, enter a unit of measurement for the attribute.

 Note

• You can define minimum and maximum values for numeric attributes to ensure valid user input.
The system will check these values during user input and displays an error message if the input
is outside the defined range. Invalid entries are blocked from submission and are also validated
before being sent to the backend or machine learning model.
• When the attribute type is set to Standard, the attribute will be displayed as a Dropdown in the
runtime application. When the attribute type is set to Extended, the attribute will be displayed as a
Value Help field, supporting more than 200 selectable values.

7. To add attributes from an existing product selection profile, choose Add from Reference and perform the
following steps.

 Note

To add attributes from reference of another product selection profile, you must already have an existing
product selection profile with the same product category.

a. In the Add from Reference dialog, choose  value help in the IPSP field.
b. Select an existing product selection profile in the IPSP dialog that opens, and choose OK.
a. Choose Add from Reference.
8. In the Product Attributes section, choose Create to build a configuration profile from scratch.

The system creates an entry for you to assign the solutions to the product attribute.

 Note

You can create multiple solutions for a product attribute.

9. In the new entry that is created, perform the following steps.

a. Choose  value help in the first field to select a solution from the Solution dialog that gets assigned to
the configuration profile.
a. Choose  value help in the second field to select a characteristic from the Characteristic dialog that
gets assigned as an attribute to the configuration profile.
b. In the third field, choose  navigation-down-arrow to select an attribute type for the assigned attribute.
c. In the last field, enter a unit of measurement for the attribute.
10. To add attributes from an existing product selection profile, choose Add from Reference and perform the
following steps

Configuring SAP Intelligent Product Recommendation

10 PUBLIC Preparing Data for Ingestion and Transformation
  Note

To add attributes from reference of another product selection profile, you must already have an existing
product selection profile with the same product category.

a. In the Add from Reference dialog, choose  value help in the IPSP field.
b. Select an existing product selection profile in the IPSP dialog that opens, and choose OK.
c. Choose  value help in the Solution field.
d. Select an existing solution in the Select: Solution dialog that opens, and choose OK.
a. Choose Add from Reference.
11. If you want to use any of the product attributes to determine the lead time, and price predictions of
recommended products, then select the checkbox in the Determines Commercial Attributes column of the
attribute.
12. To convert the price estimates to alternative currencies, you can add alternate currencies in the
Commercial Attributes Settings for Runtime Application. To do this perform the following steps.
a. Click Edit to edit the currency details.
b. Choose  value help in the Alternative Currencies field.
c. Select the alternative currencies in which you want the conversions to take place for the runtime
application and click Ok.
d. Click Save.

Product selection profile will be saved with the alternative currencies.

13. If you want to add your own custom data to a product attribute, then select the checkbox in the Additional
Product Attribute column of the attribute. This allows you to include additional information that may
not be covered by standard product attributes, helping refine recommendations and enhance decision-
making.

 Note

In case of a SAP S/4HANA scenario, once you select the Additional Product Attribute checkbox, while
preparing the data for ML you need to create an Engineered Dataset. The columns for the Additional
Product Attribute will remain empty and you can manually enter the data for these attributes.

14. Choose Create.

Task overview: Preparing Data for Ingestion and Transformation [page 7]

Previous: Assigning Products to Solutions [page 8]

Next task: Synchronizing Descriptions of Attributes and Configurable Products [page 12]

Configuring SAP Intelligent Product Recommendation

Preparing Data for Ingestion and Transformation PUBLIC 11
2.4 Synchronizing Descriptions of Attributes and
Configurable Products

Synchronize descriptions of attributes and configurable products that are associated with the IPSPs with the
source systems so that these descriptions and names can be displayed in the runtime application in different
languages.

Prerequisites

Your required product selection profile's Description Sync Status is Not Started.

Procedure

1. Launch the Manage Intelligent Product Selection Profiles application.

2. Choose Go.
3. Select the required IPSP from the Profiles table, and choose Sync Descriptions.

 Note

You can also choose Sync Descriptions in the object page of the product selection profile.

4. Choose Refresh to see the latest status in the Description Sync Status column.

Task overview: Preparing Data for Ingestion and Transformation [page 7]

Previous task: Creating Product Selection Profiles for Product Categories [page 9]

Next task: Informing Machine Learning Specialist about IPSP Activation [page 13]

Configuring SAP Intelligent Product Recommendation

12 PUBLIC Preparing Data for Ingestion and Transformation
2.5 Informing Machine Learning Specialist about IPSP
Activation

Let your machine learning specialist know about your IPSP that needs to be activated.

Context

Once you have created a product selection profile (IPSP), you must activate it to be used for machine learning
training.

 Note

You can activate a product selection profile (IPSP) only if your machine learning specialist has linked it to a
machine learning profile, and published it.

Procedure

1. Contact your machine learning specialist with the details of your IPSP so that they can associate it with a
machine learning profile, and publish it.
2. Check back with your machine specialist if you can activate your IPSP or not.

Task overview: Preparing Data for Ingestion and Transformation [page 7]

Previous task: Synchronizing Descriptions of Attributes and Configurable Products [page 12]

Configuring SAP Intelligent Product Recommendation

Preparing Data for Ingestion and Transformation PUBLIC 13
3 Preparing Data for Machine Learning

Follow the steps to prepare your data for machine learning trainings to get the best product recommendations.

1. Creating Raw Datasets [page 15]

Raw datasets act as containers for data that you import from various source systems. You can also use
these raw datasets for reference when you create engineered datasets.
2. Importing Data to Raw Dataset [page 16]
Add data to a raw dataset by importing it from a source system.
3. Creating Engineered Dataset [page 17]
Use engineered datasets to modify your downloaded data and upload it back.
4. Generating Document [page 18]
Generating document lets you define parameters of the data that you require and make it available it as
a document in SAP Intelligent Product Recommendation.
5. Downloading Document to Local System [page 19]
Downloading the generated document to your local system allows you to modify the contents of the
document, and upload it back with the changes.
6. Uploading Document [page 19]
Uploading a document lets you transfer your modified or fresh data to SAP Intelligent Product
Recommendation.
7. Importing Document [page 21]
Importing document makes your data available for machine learning trainings.
8. Creating Material Variant Datasets [page 21]
Use variant datasets to upload variants of configurable producrs.
9. Generating Document [page 22]
Generating document lets you define parameters of the data that you require and make it available it as
a document in SAP Intelligent Product Recommendation.
10. Uploading Material Variant Data [page 23]
Uploading material variant document lets you transfer your data to SAP Intelligent Product
Recommendation.
11. Importing Document [page 24]
Importing document makes your data available for machine learning trainings.
12. Viewing Runtime Dataset [page 25]
A runtime dataset contains information about a selected product and its configuration in the runtime
application.

Previous task: Preparing Data for Ingestion and Transformation [page 7]

Next task: Preparing Machine Learning Profiles for Recommendations [page 26]

Configuring SAP Intelligent Product Recommendation

14 PUBLIC Preparing Data for Machine Learning
3.1 Creating Raw Datasets

Raw datasets act as containers for data that you import from various source systems. You can also use these
raw datasets for reference when you create engineered datasets.

Context

Raw datasets contain raw or uncleansed data. Based on your requirement, you can choose to use a raw dataset
as is or, modify it and then use it for machine learning.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose Create Raw Dataset .

3. In the Create Raw Dataset dialog, enter the following information.

Field Name Description

Dataset Name of the dataset.

 Note
Ensure that the dataset name doesn't contain these
characters: equal sign (=), plus sign (+), at sign (@),
tab (0x09) and carriage return (0x0D).

Product Category The product category for the dataset.

Solution Solution for the dataset.

4. Choose Create.

Task overview: Preparing Data for Machine Learning [page 14]

Next task: Importing Data to Raw Dataset [page 16]

Configuring SAP Intelligent Product Recommendation

Preparing Data for Machine Learning PUBLIC 15
3.2 Importing Data to Raw Dataset

Add data to a raw dataset by importing it from a source system.

Context

Importing data from a source system helps in creating a dataset that is based on types and a time range.

 Note

SAP Intelligent Product Recommendation reads specific attributes such as, sales order, sales order item,
and product ID from sales transactional data to support intelligent recommendations The extracted data
is provided to SAP Intelligent Product Recommendation in a compliant way. Consent management is not
implemented for this scenario.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the raw dataset.
3. Choose Import Data.
4. In the Import Data dialog, enter the following information.

Field Name Description

Source System The system from which you want to pull in data.

Document Type There are two types of document that you can select from:
Order and Quotation.

Data Date Range The time span for which the data is to be collected.

Pricing Procedure Procedure to determine the prices of goods by taking the

various condition types into account during processing of
their sales documents.

Gross Price Condition Type The condition technique that is used to determine gross
prices of the goods from information stored in condition
records.

COGS Price Condition Type The condition technique that is used to determine direct
costs of producing the goods from information stored in
condition records.

Comment Optional

5. Choose Import Data.

 Note

In an SAP S/4HANA scenario, when extracting data using a raw dataset, if a product mapping exists
in Manage Products Mapping (MPM) application, the system will fetch the configurable product

Configuring SAP Intelligent Product Recommendation

16 PUBLIC Preparing Data for Machine Learning
 data based on the mapping; otherwise, it will attempt to retrieve data for the assigned product in
Manage Products Mapping application. In this case, an engineered dataset must be created to replace
the mapped configurable products with the products assigned in the Manage Products Mapping
application ensuring consistency for machine learning purposes.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Creating Raw Datasets [page 15]

Next task: Creating Engineered Dataset [page 17]

3.3 Creating Engineered Dataset

Use engineered datasets to modify your downloaded data and upload it back.

Context

You can use engineered datasets to modify the downloaded data based on your requirements and upload it
back. You can add to the existing data, remove from it, or perform a data cleansing on the whole data.

 Note

While creating an engineered dataset, you need to ensure that the products listed in the engineering
dataset are matching with the products in the Manage Product Categories application.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose Create Engineered Dataset .

3. In the Create Engineered Dataset dialog, enter the following information.

Field Name Description

Dataset Name of the dataset.

Configuring SAP Intelligent Product Recommendation

Preparing Data for Machine Learning PUBLIC 17
 Field Name Description

 Note
Ensure that the dataset name doesn't contain these
characters: equal sign (=), plus sign (+), at sign (@),
tab (0x09) and carriage return (0x0D).

Product Category The product category for the dataset.

IPSP Product selection profile for the dataset.

Solution Solution for the dataset.

4. Choose Create.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Importing Data to Raw Dataset [page 16]

Next task: Generating Document [page 18]

3.4 Generating Document

Generating document lets you define parameters of the data that you require and make it available it as a
document in SAP Intelligent Product Recommendation.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the engineered dataset.
3. In the Document Staging Area tab, choose Generate Document.
4. In the Generate Document dialog box, select the datasets that you want to use as references, and enter a
document name.
5. Choose Generate.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Creating Engineered Dataset [page 17]

Next task: Downloading Document to Local System [page 19]

Configuring SAP Intelligent Product Recommendation

18 PUBLIC Preparing Data for Machine Learning
3.5 Downloading Document to Local System
Downloading the generated document to your local system allows you to modify the contents of the document,
and upload it back with the changes.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the engineered dataset.
3. In the Document Staging Area tab, choose  Download to Local System to download the document
containing the required data to your local system.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Generating Document [page 18]

Next task: Uploading Document [page 19]

3.6 Uploading Document

Uploading a document lets you transfer your modified or fresh data to SAP Intelligent Product
Recommendation.

Prerequisites

• Ensure that you’ve renamed the document that you previously downloaded from the application.
• Make sure your document is virus-free.
• Ensure that the date is in the YYYYMMDD format.
• Ensure that the cell values of the file that you upload don't begin with these characters: equal sign (=), plus
sign (+), at sign (@), tab (0x09) and carriage return (0x0D).

Context

 Note

SAP Intelligent Product Recommendation reads specific attributes such as, sales order, sales order item,
and product ID from sales transactional data to support intelligent recommendations The extracted data

Configuring SAP Intelligent Product Recommendation

Preparing Data for Machine Learning PUBLIC 19
 is provided to SAP Intelligent Product Recommendation in a compliant way. Consent management is not
implemented for this scenario.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the engineered dataset.
3. In the Document Staging Area tab, choose Upload Document.
4. In the Upload Document dialog, choose  to search for the file that you want to upload.

 Caution

Avoid closing your browser during the upload process.

5. Choose Upload Document.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Downloading Document to Local System [page 19]

Next task: Importing Document [page 21]

3.6.1 Data Extraction from SAP ECC System

If you use SAP ERP Central Component (ECC) and don't have connectivity to other systems, you can still
extract data from your SAP ECC system.

You can extract data from SAP ECC by running a report. Once the data is extracted to your application
server, you can modify it before uploading, or upload it as is to the Manage Data Ingestion and Transformation
application for machine learning trainings. For more information, refer to Configuration section. The Report for
SAP ECC Data Extraction contains all the information that you need to download and run the report to extract
data from SAP ECC.

 Remember

• You need S-user ID to login, and download these reports.

• Read the Sales Item Variant Configuration Extraction Report document before you start with the
extraction process.

Configuring SAP Intelligent Product Recommendation

20 PUBLIC Preparing Data for Machine Learning
3.6.2 Lead Time and Price Prediction Calculation

Manually provide and upload lead time and price prediction information for configurable products.

If you want to calculate the lead time and price predictions of configurable products in the runtime app, then
you can manually provide the data in an engineered dataset. Choose Generate Template, fill it up with the lead
time, and pricing-related information, and choose Upload Document to upload this updated data in SAP IPR.

3.7 Importing Document

Importing document makes your data available for machine learning trainings.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the engineered dataset.
3. In the Document Staging Area tab, choose  Import to import the document containing the required data.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Uploading Document [page 19]

Next task: Creating Material Variant Datasets [page 21]

3.8 Creating Material Variant Datasets

Use variant datasets to upload variants of configurable producrs.

Context

You can use material variant datasets to upload data which is used to search for material variants and their
corresponding configurations.

Configuring SAP Intelligent Product Recommendation

Preparing Data for Machine Learning PUBLIC 21
Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose Create Material Variant Dataset .

3. In the Create Material Variant Dataset dialog, enter the following information.

Field Name Description

Dataset Name of the dataset.

 Note
Ensure that the dataset name doesn't contain these
characters: equal sign (=), plus sign (+), at sign (@),
tab (0x09) and carriage return (0x0D).

Product Category The product category for the dataset.

IPSP Product selection profile for the dataset.

Solution Solution for the dataset.

 Note

You can have only one dataset for a solution of an IPSP.

4. Choose Create.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Importing Document [page 21]

Next task: Generating Document [page 22]

3.9 Generating Document

Generating document lets you define parameters of the data that you require and make it available it as a
document in SAP Intelligent Product Recommendation.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the material variant dataset.
3. In the Document Staging Area tab, choose Generate Document.

Configuring SAP Intelligent Product Recommendation

22 PUBLIC Preparing Data for Machine Learning
 4. In the Generate Document dialog box, select the datasets that you want to use as references, and enter a
document name.
5. Choose Generate.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Creating Material Variant Datasets [page 21]

Next task: Uploading Material Variant Data [page 23]

3.10 Uploading Material Variant Data

Uploading material variant document lets you transfer your data to SAP Intelligent Product Recommendation.

Prerequisites

• Make sure your document is virus-free.

• Ensure that the date is in the YYYYMM format.
• Ensure that the cell values of the file that you upload don't begin with these characters: equal sign (=), plus
sign (+), at sign (@), tab (0x09) and carriage return (0x0D).

Context

 Note

SAP Intelligent Product Recommendation reads specific attributes such as, product ID and material variant
ID from sales transactional data to support intelligent recommendations The uploaded data is provided to
SAP Intelligent Product Recommendation in a compliant way. Consent management is not implemented
for this scenario.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the material variant dataset.
3. In the Document Staging Area tab, choose Upload Document.
4. In the Upload Document dialog, choose  to search for the file that you want to upload.

Configuring SAP Intelligent Product Recommendation

Preparing Data for Machine Learning PUBLIC 23
  Caution

Avoid closing your browser during the upload process.

5. Choose Upload Document.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Generating Document [page 22]

Next task: Importing Document [page 24]

3.11 Importing Document

Importing document makes your data available for machine learning trainings.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the material variant dataset.
3. In the Document Staging Area tab, choose  Import to import the document containing the required data.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Uploading Material Variant Data [page 23]

Next task: Viewing Runtime Dataset [page 25]

Configuring SAP Intelligent Product Recommendation

24 PUBLIC Preparing Data for Machine Learning
3.12 Viewing Runtime Dataset

A runtime dataset contains information about a selected product and its configuration in the runtime
application.

Context

This is a third type of dataset that you can’t create in the application, as its data is derived from the runtime
application only. When a user in the runtime application selects and accepts a product, then all the details
pertaining to it is stored in the runtime dataset.

Procedure

1. Launch the Manage Data Ingestion and Transformation application.

2. Choose  to go to the details page of the engineered dataset.
3. In the Document Staging Area tab, choose Generate Document.
4. In the Generate Document dialog box, select the runtime dataset, and enter a document name.
5. Choose Generate.

The system generates a document containing the runtime data.

6. Choose  Download to Local System to download the document.

Task overview: Preparing Data for Machine Learning [page 14]

Previous task: Importing Document [page 24]

Configuring SAP Intelligent Product Recommendation

Preparing Data for Machine Learning PUBLIC 25
4 Preparing Machine Learning Profiles for
Recommendations

Follow these steps to create, train, and deploy ML models to check their predictive performances, and select
the best match for your requirements.

1. Creating Machine Learning Profiles [page 26]

Machine learning (ML) profiles act as containers for machine learning models. You use these models
for machine learning trainings.
2. Creating Machine Learning Models [page 27]
Machine learning models help you to specify the response you would like to record and establish
constraints for the models to operate with.
3. Training Machine Learning Models [page 28]
Train your machine learning models to learn from patterns, and discover insights.
4. Deploying Machine Learning Models [page 30]
Deploy your machine learning models so that the predictions generated by them can be made available
to the runtime application.
5. Publishing Machine Learning Profiles [page 30]
Make machine learning recommendations available in the runtime application by publishing them to
help a guided selling expert choose the best product for their business requirements.
6. Informing Product Manager About Published ML Profile [page 31]
Let your product manager know about the machine learning profile being published so that they can
activate the IPSP.

Previous task: Preparing Data for Machine Learning [page 14]

Next task: Activating Product Selection Profiles (IPSP) for Product Recommendations [page 32]

4.1 Creating Machine Learning Profiles

Machine learning (ML) profiles act as containers for machine learning models. You use these models for
machine learning trainings.

Procedure

1. Launch the Manage Machine Learning Lifecyle application.

2. Choose Create.

Configuring SAP Intelligent Product Recommendation

26 PUBLIC Preparing Machine Learning Profiles for Recommendations
3. In the Create dialog, enter the following information.

Field Name Description

ML Profile Name of the machine learning profile

Description Description of the profile.

Product Category Product category that needs to be associated with the ML

profile.

IPSP IPSP that needs to be associated with the ML profile. The

IPSP contains the attributes, and characteristics of the
desired product and its configuration.

 Note

Ensure that the ML profile name and description don't contain these characters: equal sign (=), plus
sign (+), at sign (@), tab (0x09) and carriage return (0x0D).

Task overview: Preparing Machine Learning Profiles for Recommendations [page 26]

Next task: Creating Machine Learning Models [page 27]

4.2 Creating Machine Learning Models

Machine learning models help you to specify the response you would like to record and establish constraints for
the models to operate with.

Context

ML models use algorithms to learn from. You can get useful insights and predictions using an ML model.
Solution targets can have two types of ML models. They are:

• Configuration: This type of ML model helps in determining the best composition and make of the
recommended products.
• Commerical Attribute: This type of ML model helps in determining the lead time, net price, gross price, and
Cost of Goods Sold (COGS) of the recommended products.

Procedure

1. Launch the Manage Machine Learning Lifecycle application.

2. Choose  to go to the details page of your required machine learning profile.

The ML profile page opens, and you can see two targets in the Determination Model Overviews.

Configuring SAP Intelligent Product Recommendation

Preparing Machine Learning Profiles for Recommendations PUBLIC 27
  Note

There can be only two types targets in a machine learning profile. One is a product category and the
other is a solution.

3. Choose  to go to the details page of your required target.

4. In the details page that opens, choose Create and enter a name, and description of the ML model.

 Note

• Ensure that the ML profile name and description don't contain these characters: equal sign (=),
plus sign (+), at sign (@), tab (0x09) and carriage return (0x0D).
• When you create an ML model for a solution target, you see Type field. Choose the required type
from the Type dropdown list.

5. Select datasets that you require for the ML model in the Dataset dropdown list.

The system creates a new entry.

 Note

• You can create multiple machine learning models for a target.

• For a product category target, you can set a value to determine the number of products to be
displayed in the runtime app in Maximum Results to be Displayed, and specify the threshold value
in Lower Threshold in the Product Display Configuration at Runtime screen area.
• In case of a S/4HANA scenario, where Additional Product Attributes are used, ensure you use an
Engineered Dataset for ML training.

Task overview: Preparing Machine Learning Profiles for Recommendations [page 26]

Previous task: Creating Machine Learning Profiles [page 26]

Next task: Training Machine Learning Models [page 28]

4.3 Training Machine Learning Models

Train your machine learning models to learn from patterns, and discover insights.

Procedure

1. Launch the Manage Machine Learning Lifecycle application.

2. Choose  to go to the details page of your required machine learning profile.
3. In the Determination Model Overviews table, choose  to go to the details page of your required target.

Configuring SAP Intelligent Product Recommendation

28 PUBLIC Preparing Machine Learning Profiles for Recommendations
4. In the <target name> table, select your required machine learning model.

The ML model that you want to train must have ML Model Status as Training File Generated, Ready for
Training.
5. Choose Train.

The status of the ML model changes to Training.

Task overview: Preparing Machine Learning Profiles for Recommendations [page 26]

Previous task: Creating Machine Learning Models [page 27]

Next task: Deploying Machine Learning Models [page 30]

4.3.1 The Best Machine Learning Models for

Recommendations

Identify and deploy ML models that are best-suited for your product recommendations based on your
requirements.

The two types of ML models supported by SAP Intelligent Product Recommendation along with the associated
metrics are mentioned below.

You can decide on your choice of ML model to deploy based on these metrics.

Model Type Supported Metrics

Configuration model
 Note
All three metrics range from 0 to 1 with 1 being most
accurate and 0 being least accurate.

• Precision - ability of the model to assign only the

relevant records to the correct attributes
• Recall - ability of the model to find all the records
belonging to the correct attributes
• FI Score - harmonic mean of precision and recall

Commercial Attribute model • Mean Absolute Error (MAE) - average absolute

difference between the predicted values and the actual
value. The lower the better.
• Mean Absolute Percentage Error (MAPE) - average of
the absolute percentage errors of the predictions. The
lower the better.
• Mean Squared Error (MSE) - average squared
difference between the predicted values and the actual
value. The lower the better.

Configuring SAP Intelligent Product Recommendation

Preparing Machine Learning Profiles for Recommendations PUBLIC 29
4.4 Deploying Machine Learning Models

Deploy your machine learning models so that the predictions generated by them can be made available to the
runtime application.

Procedure

1. Launch the Manage Machine Learning Lifecycle application.

2. Choose  to go to the details page of your required machine learning profile.
3. In the Determination Model Overviews table, choose  to go to the details page of your required target.
4. In the <target name> table, select your required machine learning model.

The ML model that you want to deploy must have ML Model Status as Trained.
5. Choose Deploy.

 Note

The system can take some time to deploy the ML model. In such cases, select the ML model and
choose Refresh to view the latest status.

The status of the ML model changes to Deployed.

Task overview: Preparing Machine Learning Profiles for Recommendations [page 26]

Previous task: Training Machine Learning Models [page 28]

Next task: Publishing Machine Learning Profiles [page 30]

4.5 Publishing Machine Learning Profiles

Make machine learning recommendations available in the runtime application by publishing them to help a
guided selling expert choose the best product for their business requirements.

Procedure

1. Launch the Manage Machine Learning Lifecycle application.

2. In the Machine Learning Profiles table, select an ML profile that has deployed models.

Configuring SAP Intelligent Product Recommendation

30 PUBLIC Preparing Machine Learning Profiles for Recommendations
  Note

The ML profile must have at least one deployed model for all the targets.

3. Choose Publish.

 Note

If you want to unpublish an ML profile, you must publish another ML profile with the same IPSP in its
place.

Task overview: Preparing Machine Learning Profiles for Recommendations [page 26]

Previous task: Deploying Machine Learning Models [page 30]

Next task: Informing Product Manager About Published ML Profile [page 31]

4.6 Informing Product Manager About Published ML Profile

Let your product manager know about the machine learning profile being published so that they can activate
the IPSP.

Procedure

Contact your product manager with the details of your machine learning profile that is linked to the required
IPSP.

Task overview: Preparing Machine Learning Profiles for Recommendations [page 26]

Previous task: Publishing Machine Learning Profiles [page 30]

Configuring SAP Intelligent Product Recommendation

Preparing Machine Learning Profiles for Recommendations PUBLIC 31
5 Activating Product Selection Profiles
(IPSP) for Product Recommendations

Activate a product selection profile (IPSP) so that it can be used for recommending appropriate products in the
runtime application.

Prerequisites

Ensure that your machine learning specialist has already associated your IPSP with a published machine
learning profile.

Procedure

1. Launch the Manage Intelligent Product Selection Profiles application.

2. Choose Go.
3. Select the product selection profile that you want to activate from the table.
4. Choose Activate.

Previous task: Preparing Machine Learning Profiles for Recommendations [page 26]

Next: Built-in Support for SAP Intelligent Product Recommendation [page 33]

5.1 Product Selection Profile (IPSP) Deactivation

You can deactivate an active IPSP. However, this deactivation also unpublishes the ML profile that is associated
with the IPSP and undeploys its ML models.

To deactivate an IPSP, select it from the table and choose Deactivate.

Configuring SAP Intelligent Product Recommendation

32 PUBLIC Activating Product Selection Profiles (IPSP) for Product Recommendations
6 Built-in Support for SAP Intelligent
Product Recommendation

SAP Intelligent Product Recommendation leverages the capabilities of Built-In Support to provide embedded
support with direct and context aware access to support knowledge base, and support channels such as,
support incidents or expert chat, without the need to leave the application.

Previous task: Activating Product Selection Profiles (IPSP) for Product Recommendations [page 32]

Related Information

Accessing Built-In Support Application

Report Issue to SAP

Configuring SAP Intelligent Product Recommendation

Built-in Support for SAP Intelligent Product Recommendation PUBLIC 33
Important Disclaimers and Legal Information

Hyperlinks
Some links are classified by an icon and/or a mouseover text. These links provide additional information.
About the icons:

• Links with the icon : You are entering a Web site that is not hosted by SAP. By using such links, you agree (unless expressly stated otherwise in your
agreements with SAP) to this:

• The content of the linked-to site is not SAP documentation. You may not infer any product claims against SAP based on this information.

• SAP does not agree or disagree with the content on the linked-to site, nor does SAP warrant the availability and correctness. SAP shall not be liable for any
damages caused by the use of such content unless damages have been caused by SAP's gross negligence or willful misconduct.

• Links with the icon : You are leaving the documentation for that particular SAP product or service and are entering an SAP-hosted Web site. By using
such links, you agree that (unless expressly stated otherwise in your agreements with SAP) you may not infer any product claims against SAP based on this
information.

Videos Hosted on External Platforms

Some videos may point to third-party video hosting platforms. SAP cannot guarantee the future availability of videos stored on these platforms. Furthermore, any
advertisements or other content hosted on these platforms (for example, suggested videos or by navigating to other videos hosted on the same site), are not within
the control or responsibility of SAP.

Beta and Other Experimental Features

Experimental features are not part of the officially delivered scope that SAP guarantees for future releases. This means that experimental features may be changed by
SAP at any time for any reason without notice. Experimental features are not for productive use. You may not demonstrate, test, examine, evaluate or otherwise use
the experimental features in a live operating environment or with data that has not been sufficiently backed up.
The purpose of experimental features is to get feedback early on, allowing customers and partners to influence the future product accordingly. By providing your
feedback (e.g. in the SAP Community), you accept that intellectual property rights of the contributions or derivative works shall remain the exclusive property of SAP.

Example Code
Any software coding and/or code snippets are examples. They are not for productive use. The example code is only intended to better explain and visualize the syntax
and phrasing rules. SAP does not warrant the correctness and completeness of the example code. SAP shall not be liable for errors or damages caused by the use of
example code unless damages have been caused by SAP's gross negligence or willful misconduct.

Bias-Free Language
SAP supports a culture of diversity and inclusion. Whenever possible, we use unbiased language in our documentation to refer to people of all cultures, ethnicities,
genders, and abilities.

Configuring SAP Intelligent Product Recommendation

34 PUBLIC Important Disclaimers and Legal Information
Configuring SAP Intelligent Product Recommendation
Important Disclaimers and Legal Information PUBLIC 35
[Link]/contactsap

© 2025 SAP SE or an SAP affiliate company. All rights reserved.

No part of this publication may be reproduced or transmitted in any form

or for any purpose without the express permission of SAP SE or an SAP
affiliate company. The information contained herein may be changed
without prior notice.

Some software products marketed by SAP SE and its distributors

contain proprietary software components of other software vendors.
National product specifications may vary.

These materials are provided by SAP SE or an SAP affiliate company for

informational purposes only, without representation or warranty of any
kind, and SAP or its affiliated companies shall not be liable for errors or
omissions with respect to the materials. The only warranties for SAP or
SAP affiliate company products and services are those that are set forth
in the express warranty statements accompanying such products and
services, if any. Nothing herein should be construed as constituting an
additional warranty.

SAP and other SAP products and services mentioned herein as well as
their respective logos are trademarks or registered trademarks of SAP
SE (or an SAP affiliate company) in Germany and other countries. All
other product and service names mentioned are the trademarks of their
respective companies.

Please see [Link] for

additional trademark information and notices.

THE BEST RUN