Hi everyone,
we have 2 independent SAP Commerce systems that share one instance of SAP Marketing Cloud (SMC). Both of them replicate their products into this instance. The first Commerce system does this using the standard product origin "SAP_HYBRIS_PRODUCT". The second Commerce system uses a custom product origin "EXAMPLE_ORIGIN" instead so we can differentiate the products in SMC.
So we have products with origin SAP_HYBRIS_PRODUCT and products with origin EXAMPLE_ORIGIN
Note: Some products are present in both shops, but differ in some attribute values (for example product url, since its a different shop)
This was working quite good for filtering products of only one shop - for example in segmentation. But we are now facing a problem with recommendations. We have the same recommendation scenario for both shops so we created a recommendation model typ in SMC for each shop:
For both recommendation model types a corresponding recommendation scenario was created with its own ID. Both scenarios are working and recommend products.
So what is the problem?
In "Manage Recommendations" app in SMC you can see statistics like impressions and click rate for your scenario. This is working for SHOP_A_TOP_SELLER with the recommendation type "SAP Commerce:Product" but not for SHOP_B_TOP_SELLER with the custom type. Statistics are only shown for the standard type. If we change SHOP_B_TOP_SELLER to the type of SHOP_A_TOP_SELLER we can see the statistics.But then our recommendation does not consider leading items with product origin "EXAMPLE_ORIGIN" any more and returns an empty result.
Additionally we noticed that the recommendations return products from both product origins. Our idea was to change the origin "EXAMPLE_ORIGIN" to the standard origin as well and tell them apart by a custom field, but this would not work for products that are present in both shops, since they would overwrite each other all the time.
So in short: We want to have 2 product origins. We want our recommendations to only consider a specific origin (the standard one or the custom one - depending on the shop the scenario is created for) and to see the statistics of recommendations with our custom origin as well. How is this done? Do we maybe have a conceptual error?
Thank you,
Sebastian