SAP Marketing Cloud Recommendation Algorithm "Often Bought Together"

[hackbu]

We have integrated the SAP Marketing Cloud with Commerce and are trying to use recommendations now. We decided to choose the "Often Bought Together (Interactions)" algorithm to determine recommendations when a user has opened a product detail page.

We have SALES_ORDER interactions available in the Marketing Cloud. The origin of the products is SAP_HYBRIS_PRODUCT.

I have some question regarding the selected algorithm: Does this consider the orders of all customers or just the ones of the customer? So in case the customer which is currently visiting the product page did never purchase this product before, will he still get some recommendation if other customers already purchased the product together with other products?

Do we need to make additional adaptions (to the algorithm) for this setup since we are working with SAP_HYBRIS_PRODUCT?

[SCHNEIDERT · Accepted solution]

Hi Christian,

it consideres all of the interactions in the given time frame that you set (by default last 60 days).

So yes, by default the Customer will also get the recommendations if other Customers bought the product together with other products.

BR Tobias

[hackbu · Accepted solution]

Hey schneidert,

thanks for the feedback again. I also updated the question. The orders work with products of type SAP_HYBRIS_PRODUCT. I think this is "SAP Commerce Product":

Does the algorithm automatically work with this or do we need to make some additional change?

Because the description of the algorithm is:

"This algorithm uses records from the following data sources:

• Transactions: SAP Marketing Interactions - (CEI_IA_IC)
• Items: SAP Marketing Product - (CEI_CUAN_PRODUCT)
• Users: SAP Marketing Interaction Contact - (CEI_IC_CONSUMER)

The records are grouped by session ID using SQL (one leading item)."

For the items it says "SAP Marketing Product" - I'm not sure if this is the same...

BR
Christian

[SCHNEIDERT · Accepted solution]

Hey hackbu,

by default the model uses the SAP Marketing Product.

You have to adjust this in the "Recommendation Model Types" app.

Choose the one which your Recommendation Model is based on and change the Recommendation Type to SAP Commerce Product (which is SAP_HYBRIS_PRODUCT as you already pointed out).

BR Tobias

[hackbu · Accepted solution]

Hey schneidert,

thanks for the explanation. We just did this. I think we had one additional issue which led to bad recommendation data. The algorithm by default did not consider interaction contacts of type company which in our case are about 3/4 of all contacts. I adapted the algorithm filter and now it looks better. Thanks for the help!

BR
Christian

[SCHNEIDERT · Accepted solution]

Hey Christian,

good to hear that it works! You're welcome 🙂

BR Tobias