The Einstein Vision model is a representation of how objects are detected and counted in an image. The model needs to be created as part of the development process using a base set of images. The base set of images should contain images of the products that need to be detected on the shelf, as well as labels and annotations for each product. The model is then trained and tested using these images and can be improved by adding more images or adjusting the parameters. Verified References: [Salesforce Consumer Goods Cloud Implementation Guide], page 23-24.

For partner-facing guided journeys in Experience Cloud that must communicate with external systems during order capture, Salesforce recommends using OmniScripts and Integration Procedures together.

OmniScripts (A) provide the guided user journey —highly configurable, step-by-step flows where partners can configure products, enter customer data, and trigger backend processes. OmniScripts are the front-end orchestration layers for Communications Cloud CPQ and order capture in portals.

Integration Procedures (C) are the server-side execution engine that handle:

Calling external systems

Orchestrating multiple APIs

Handling transformations

Reducing client-server chattiness

They are specifically optimized for Experience Cloud, where minimizing round trips and improving performance is critical.

DataRaptors (B) handle simple CRUD and transformations but are not well suited for multi-step external integrations.

FlexCards (D) are for display/UI components, not guided journeys or external API orchestration.

A promotion check is a type of action plan task that allows a field rep to verify if a promotion is running correctly at a retail store. When executing a promotion check in the field, two standard capabilities are available: viewing which products are included in the promotion and taking a photo of the promotion. These capabilities help to ensure compliance and capture evidence of promotional execution. Verified References: [Salesforce Consumer Goods Cloud Implementation Guide], page 27-28.

This approach allows for an efficient and code-minimal way to check compliance of high-value assets in stores and trigger correction action plans based on Boolean values.

These dashboards provide comprehensive insights into various aspects of consumer goods operations, from category performance to individual sales rep and store analyses.

Store Action Plan Templates for recommended tasks can be defined at various levels including Retail Store Group, Retail Store, and In-Store Location, allowing for granular and effective task management.

These steps are essential in setting up Einstein Vision for automatic product detection on shelves, involving enabling the feature, defining relevant tasks, and training the models with appropriate product images.

In Retail Store KPIs, you can define the assessment indicator definition that must be captured for a store group. Because the assessment indicator definition is associated with assessment task definition and store groups to stores, Retail Store KPIs are the link between a store and the tasks to be performed at a store. You can also define the expected values of an assessment indicator definition in Retail Store KPIs

In Salesforce Communications Cloud and Industries Order Management (OM), products that do not carry commercial value but are required for fulfillment must be modeled as Technical Products . Public Salesforce EPC and OM design guidance explains that a Technical Product is used when an item is not visible during selling , does not influence the commercial configuration, and is instead required for downstream fulfillment , service activation , or device provisioning .

In the case of Universal Connect (UC), the router included with the Business Internet offer has no commercial significance —meaning sales agents do not configure it, no pricing is associated with it, and customers should not see it as a selectable or independent commercial item. Instead, it simply needs to appear in the fulfillment layer so that the order can drive device shipment , installation , inventory allocation , or activation steps .

Salesforce’s recommended modeling approach is to:

Create the router as a Technical Product Specification .

Add it to the Business Internet offer as a Technical Child .

Allow decomposition in OM to generate Fulfillment Request Lines (FRLs) for the router automatically, without exposing it during quoting.

This ensures:

Clean commercial experience (router hidden from the sales journey).

Accurate fulfillment instructions to downstream systems.

No need for complex rules or hiding logic in quote documents.

Options A and B incorrectly place the router inside the commercial structure, which is not aligned with EPC best practices. Option D introduces a separate offer and automation logic, which adds unnecessary complexity and breaks modeling principles.