This online publication is intellectual property of Apptus Technologies. Its contents can be duplicated in part or whole, provided that a copyright label is visibly located on each copy and the copy is used in conjunction with the product described within this document.
All information found in these documents has been compiled with utmost attention to detail. However, this does not guarantee complete accuracy. Neither Apptus Technologies nor the authors shall be held liable for possible errors or the consequences thereof.
Software and hardware descriptions cited in these documents might be registered trademarks. All trade names are subject to copyright restrictions and may be registered trademarks. Apptus Technologies essentially adheres to the manufacturer’s spelling. Names of products and trademarks appearing in this document, with or without specific notation, are likewise subject to trademark and trade protection laws and may thus fall under copyright restrictions.CLOSE
Working with Recommendations¶
Recommending relevant products requires context, and what type of panel to use for recommendations and where is up to the retailer. Recommendations with Apptus eSales Enterprise are made at the product level and products are returned according to relevance with variants returned according to a best selling order.
Selecting the right panel to use for product recommendations can be expressed as a conditional flowchart. A retailer can by answering a few questions get an idea of what panel to use. Depending on if the area where the panel is used is cached or not restricts what panels can be used.
Caching eSales content
Caching of eSales content is generally not recommended. Contact Apptus Support for more information.
|Panel||When to use|
|Recently viewed panel||If there is a need for personalised recommendations that are not predictive or actively support product retargeting. Can not be used in cached areas. Must have backfill for visitors with no previous data.|
|Abandoned carts panel||If there is a need for personalised recommendations that are not predictive while actively supporting product retargeting. Can not be used in cached areas. Must have backfill for visitors with no previous data.|
|Recommend based on customer panel||If there is a need for personalised recommendations that are predictive. Can not be used in cached areas. Must have backfill for visitors with no previous data.|
|Recommend based on cart panel||If there is a need to recommend specific products based on multiple products. Recommendations will not be personalised. Can be used in cached areas.|
|Recommend based on product panel||If there is a need to recommend specific products based on a single product. Recommendations will not be personalised. Can be used in cached areas. Active promotions and demotions can affect the recommended products.|
|Top sellers panel||If there is no need to recommend specific products, the Top sellers panel can be used for predictive recommendations within a category or with a specific product attribute using a filter argument. Recommendations will not be personalised. Can be used in cached areas.|
Fine tuning of the panels used for product recommendations can be done with additional product filtering, deduplication settings, and testing with competing panels.
The panels used for product recommendations can all use a filter argument either in a panel query or as a configured argument. The filter is an expression that limits the product and variant set used for the recommendations. This can be used to exclude products that are out of stock or on sale.
Deduplication is the concept of automatically removing duplicate products from panels in zones. The duplicate product can either be removed from the panel used to display recommendations or another panel, as long as they are within the same zone.
Two panels can be tested against each other to help additional fine tuning if the panels are in the same zone. The sub-panel setting of the zone should be either Optimized sub panel strategy or Panel split test. Once the test is concluded and there is enough data to determine a winner, stop the test and set the winning panel. Notifications for the selected panels must be correctly set up.