Product recommendations are enabled for the following POS scenarios. They are available in Cloud POS or Modern POS (MPOS).
- On the Product details page:
- If a store associate visits a Product details page when looking at previous transactions across different channels, the recommendation engine suggests additional items that are likely to be purchased together.
- If the store associate adds a customer to the transaction and then visits a Product details page, the recommendation engine provides personalized recommendations using the customer’s transaction history.
- On the Transaction page:
- The recommendation engine suggests items based on the entire list of items in the basket.
- If the store associate adds a customer to the transaction, the recommendation engine provides personal recommendations using the customer’s transaction history and the list of items in the basket.
Note To display recommendations on the Transaction page, the retailer needs to update the screen layout in Dynamics 365 for Operations. The Recommendations control must be dropped on to the Transaction page.
- On the Customer details page:
- The recommendation engine suggests items based on the user ID and items in the customer’s wish list.
To set up product recommendations, you need to do the following.
- Make sure that you have selected the correct Legal entity.
- Navigate to Entity store, select Retail sales, and then click Refresh. This will use the demo data (or your data) from your operational database and move it to Entity store.
- Optional: To display recommendations on the transaction screen, go to Screen Layout, choose your screen layout, launch the Screen layout designer, and then drop the recommendations control where needed.
- Go to Retail parameters, select Machine-learning, select Yes under Enable POS recommendations.
- To see recommendations on POS, run global configuration job 1110. To reflect changes made to POS screen layout designer, run channel configuration job 1070.
When you refresh the Entity store entity, the following actions take place.
- Data in the format required by the Cognitive services is extracted from the Dynamics 365 for Operations operational database and sent to the Entity store.
- The data is used by Azure Data Factory (ADF) to cleanse the data using Hive scripts as part of ADF activities. Cleansed data is stored in blob storage.
- Data from blob storage is used by the Cognitive services API to train a recommendation model.
When you turn on Enable recommendations and run the configuration jobs, the following actions take place.
- Model credentials and ID are picked up from the API and stored in the Dynamics 365 for Operations operational database, in the web.config for AOS, and also in the retail server.
- Model credentials and ID are made available to CRT so that calls for product recommendations from Cloud POS and MPOS in online mode can be honored.