Predictive Content Recommendation

Business Case:

KPI to be measured

Bookmark this resource Follow

Ask a question

Was this article helpful? 0 out of 0 found this helpful



Predictive Content Recommendation is a THRON feature included in the Content Intelligence module that allows you to:

  • Customize your web pages and applications with dynamic content, automatically chosen according the interests of each individual contact.
  • Collect strategic data on content, understanding how end users use it.
  • Eliminate all guesswork regarding content or asset production strategy, providing precise and concrete data to drive your choices.
  • Reduce the continuous work of updating related content, completely automating the rules with which they are proposed during the user's navigation path

The fields of application of Predictive Content Recommendation are numerous: from e-commerce to content marketing channels, from mobile applications to integration with CRM and Marketing Automation: in all these cases it is possible to increase user engagement by providing content that is always fresh and in line with their profile.

Predictive Content Recommendation offers two different types of recommendation: one is based on simple html widgets (list or wall) to be included in your pages, to propose to your users the most relevant pages (URL content) to visit, according to their interests; the second type is provided directly by the Player: at the end of the display of a content, THRON Artificial Intelligence will automatically select another content to be proposed on the basis of the user's interests.



What’s in it


During the process of creating a new Predictive Content Recommendation, you must specify:

  • The name of that specific recommendation
  • The folder on which to install it: you must choose a root folder, i.e. existing at the first level of the tree.
  • The recommendation type: via HTML5 widget or via THRON Universal Player.

Other additional parameters allow you to further refine the recommendation of content:

  • You can filter the set of content to be recommended by selecting folders
  • You can decide to penalize obsolete content
  • You can boost specific content according to their tags
  • You can decide whether or not to propose content that users have already seen

Once you have specified these parameters, you can include recommendation widgets in your projects with a simple copy&paste, or use your configuration within a specific content sharing via Universal Player. Finally, Predictive Content Recommendation provides a set of API to customize your own experience.



How to include it

Once activated, Predictive Content Recommendation can be included in web pages with any of the following methods.


Via embeddable widget


HTML5 widgets are embed codes generated automatically at the end of the configuration. They are available in two standard layouts, i.e. Wall and Horizontal List

Widgets of this type use only URL content (link to magic site article?) as they are a component designed to build dynamic navigation menus, such as content related to the same blog article or similar products in an ecommerce.

Widgets are customizable through a series of advanced parameters described here.


Via Universal Player


With this mode you can recommend any type of multimedia content: at the end of the playback of a video, audio or playlist content, a new content will be automatically proposed to the user. If you use Universal Player to embed an image in your projects, you can recommend new content after a certain period of user inactivity. To enable this type of recommendation you must set a specific flag which can be found in the embed code generation screen and then select the specific configuration of Predictive Content Recommendation to be used.



For any integration need that is not covered by THRON's widgets or Universal Player, you can rely on APIs to leverage the recommendation engine in your custom projects.

All information about the APIs to be used can be found here.

Was this article helpful?
0 out of 0 found this helpful

Have any question?

Open a ticket