Delivering Dynamic, Personalized Experiences with Adobe Target’s New User-Based Recommendations Algorithm

Your customers have more choices available to them than they’ve ever had before. Their expectations for a personalized experience are set high by personalized experiences delivered by companies like Netflix, YouTube, Spotify and Amazon. As a marketer trying to create engaging digital experiences, you need to personalize your customers’ interactions with the most relevant products, content or services, and dynamically update those offerings in real-time as users move through their journeys.

Adobe Target Recommendations’ new User-Based Recommendations algorithm allows you to deliver an individually personalized “Recommended For You” experience to each of your customers that is tailored to their unique behavior. These recommendations go beyond single-key approaches such as “favorite category”, “last viewed brand” or “people who viewed this, also viewed”, and instead use the totality of a customer’s behavior to infer their underlying preferences. For example, a user browsing a purple scarf in the winter may receive a recommendation for gloves in a complementary color; when the user returns in the spring and browses running shoes, recommendations update in-session to reflect the new interest, displaying a tank top – but may retain knowledge of the user’s preference for purple.

https://video.tv.adobe.com/v/29322

Our new algorithm is powered by Adobe Sensei and builds upon machine learning models developed by Adobe Research to power recommendations on Behance, Adobe’s discovery platform for creative work. Our approach differs from previous options by:

• Using a “keyless” approach that leverages all of the user’s recent viewing and purchasing behavior: you don’t need to base recommendations on a single item
• Recommendations are updated in-session and change as the user browses your site
• Users returning to your site will see recommendations based on their previous visit, but recommendations will change to reflect their most recent behavior

The result of our Data Science and Engineering teams’ work is an algorithm that analyzes billions of customer interactions with millions of pieces of content to create personalized recommendations for your customers, no matter what your business is.

Getting started is easy for Adobe Target Premium customers who have implemented Target Recommendations. Just add the Recommended For You criteria to one of your pages:

Or create a new Criteria and select the User-Based Recommendations option for your Recommendation Logic after selecting any item-based Recommendations Key choice (Last Viewed Item, Last Purchased Item, Current Item, or Most Viewed Item):

The selected Key doesn’t drive the User-Based Recommendations algorithm, but allows you to use Adobe Target Recommendations’ filtering options, so you can apply business logic and context to ensure a great customer experience.

Ideas and tips to drive results

Retailers

• Personalize product recommendations on your homepage by adding User-Based Recommendations, next to a Recently Viewed Items tray to help users get back to previously explored items. (Leave the “Allow Recently Viewed Items to be Recommended” option turned off to prevent duplication across trays. Consider turning on backups or using this algorithm as the first in a criteria sequence to ensure recommendations are available for users that haven’t viewed any items.)
• Encourage upsell and drive increased AOV by adding User-Based Recommendations to cart pages and/or checkout flows. (Pass excludedIds to Adobe Target to prevent items in the cart from being recommended on cart pages, and leave the “Allow Recently Purchased Items to be Recommended” option turned off for any pages shown after checkout is completed.)
• On product detail pages, “People Who Viewed This, Viewed That”, “Viewed This, Bought That”, and “Bought This, Bought That” remain the first-line options to recommend alternatives and complementary items, but User-Based Recommendations may be a useful complement.**
  **

Publishing & media

• Personalize content recommendations on your homepage by adding User-Based Recommendations to your homepage. (Consider turning on backups or using this algorithm as the first in a criteria sequence to ensure recommendations are available for users that haven’t viewed any items.)
• Drive page views and engagement by adding User-Based Recommendations on your article or item view pages next to “People Who Viewed This, Viewed That” content.

B2B, lead gen, financial services & telecom

• Personalize content recommendations on your homepages or landing pages by adding User-Based Recommendations to your homepage. (Consider turning on backups or using this algorithm as the first in a criteria sequence to ensure recommendations are available for users that haven’t viewed any items.)
• Drive page views and engagement by adding User-Based Recommendations on your article or item view pages next to “People Who Viewed This, Viewed That” content.
• Reduce the cost of service by using User-Based Recommendations on your support pages to drive visitors to the right help articles.