Personalization Engine Overview

Personalization Engine collects user behavior and interest data to recommend optimal content–your articles or products–to your users when they engage with your emails, website, and mobile app.

If you have 100 applicable content items and just seven slots to populate in an email or a section of your site, there are many trillions of possible content combinations. The system makes sure that each individual gets the most relevant combination with every delivery, driving more page views for publishers and increased average order value for retailers.

Note: If there is not enough content to meet the personalization and zephyr settings, Marigold Engage by Sailthruu will fall back to recommending content using the popular algorithm, ignoring any filtering rules and cancel/assert statements. You can also choose to manually specify and order certain content slots using Recommendation Manager, or filter content by tag to include only a certain type of content within a given context.

Here are some of the key components that make this possible:

Your Content User Profiles Recommendation Algorithms

Marigold Engage by Sailthru will maintain a copy of metadata about all of your site’s content, including item names/titles, images, URLs, prices, tags, and publication dates.

You can keep this information up to date using Google Product Sync (recommended), the Content API or the onsite JavaScript tag.

Setup:

  • Learn more about content
  • Configure Google Product Sync, Content API or JavaScript to collect updated site metadata
  • Ensure content meta tags are in place on site if the JS will be configured to collect them
As users interact with your site, the JavaScript logs their activity and attributes interests to their profile based on the tags you’ve included on your content. For each interest, we record the user’s frequency of interactions with that interest and their degree of interest compared to your overall user base.

If your site offers e-commerce features, you can also log each purchase, further augmenting user profiles.

Setup:

We have several algorithms that you can use to recommend the optimal piece of content to individual users.

Each algorithm uses different sets of data; interest tags, browse behavior, purchase behavior, additional content metadata, and more.

Setup: