#ExploreHybris – Recommendation Models in SAP Hybris Marketing
In the #ExploreHybris series, experts from Apollogic share their knowledge concerning the possibilities of the most efficient use and extending the SAP Hybris Marketing system. It is a comprehensive Marketing Automation platform used for managing marketing activities and relations with Customers.
From this article you will learn:
- How to take advantage of SAP Hybris Marketing recommendations in practice?
- How to create product recommendations?
- How do Customers behaviours impact recommended content?
Practical use of the recommendation module in SAP Hybris Marketing
The SAP Hybris Marketing recommendation module is a tool allowing to create personalized product recommendations adjusted to a specific user. This is possible thanks to taking advantage of various communication channels such as e-commerce systems, websites, social media, and e-mail campaigns.
Creating product recommendations
Creating and managing recommendation models takes place with the use of a group of dedicated SAP Fiori apps, located on the user’s Launchpad. They allow to both use the predefined models available in the system, as well as building own comprehensive recommendation models. They include for example selecting proper algorithms and defining the type and scope of the analyzed information.
Benefiting from the recommendation module allows the user to precisely adjust the recommended items, taking into consideration their popularity, mutual search frequency, or the activity of users with the most similar preferences.
- Presenting Customers content adjusted to their preferences
- Immediate response to specific needs
- Selecting products basing on external conditions, like weather and location
- Taking advantage of current Social Media trends (real-time marketing)
- Evaluating the campaign’s efficiency basing on conversion
Apart from the recommendation layer, which constitutes the base of the model, the user gains the possibility to flexibly adjust it by using layers for changing the ranking of recommended positions and filtering returning results.
In the example presented below, we have used the system’s standard algorithm which creates recommendations based on the most frequently displayed common products. We’ve additionally enriched the model with a unique algorithm created by us, which changes the rankings of positions possessing the same accuracy, basing on their pricing with the exclusion of products already in the user’s cart.
The simple interface allows also to define additional options, such as the model’s refreshing frequency and limiting the number of returns on each stage of results.
After designing the model it is possible to perform a simulation of product recommendations. Thanks to this, the user may verify whether the model works in accordance with the assumptions, before using it for example in an e-commerce system.
The tool allows performing a simulation, introducing to the model such parameters which include:
- The user,
- Currently viewed product,
- Products located in the cart and context parameters
Depending on the prepared scenario, the algorithms may require adding various numbers of attributes.
In the presented example we have added the currently viewed item and a product previously located in the viewer’s cart.
The goal was to simulate a situation in which the user of an online store browses through a website offering coffee beans, already having a mug in the cart.
Its effect is a recommended list of products which users viewed together with coffee beans most often, excluding a mug already located in the cart. Additionally, items with an identical accuracy level have been ordered accordingly to their price – in accordance with our assumption that black tea is cheaper than green tea.
As it can be seen, the interface for designing recommendation models proves to be easy and intuitive to use. It allows not only to quickly create, but also to test recommendation models.
Are you interested in SAP Hybris Marketing platform? Visit our website – SAP Hybris Marketing in a nutshell!
- On 16/10/2017