Share on facebook
Share on twitter
Share on linkedin
Share on email

Froomle’s solution to the shared account challenge

Several AI personalization engines cannot distinguish between multiple users within a single profile. Why? Recommendations are based on a consumer’s viewing and buying history; once multiple users use a single profile, the reference pool is muddied with influences outside of the original user’s interests.

THE CHALLENGE
DEVELOP AN AI PERSONALISATION ENGINE THAT RECOGNISES
MULTIPLE USERS WITHIN THE SAME PROFILE.

An AI engine analyses a consumer’s viewing and buying history to make recommendations on what they might like. The more data the engine has to work with, the better it becomes at predicting your likes and dislikes. But, what happens when multiple users use a single profile? The engine receives more data, but the reference pool is now muddied with influences outside of the original user’s interests. A simpler AI engine won’t dier between these interfering points of reference and will try its best to combine said interests.

The result: At best, if your interests don’t vary too much, you’ll still receive something you might like. At worst, your interests vary so much that the engine will try to create a mash-up. In the absence of the existence of that style, genre, or item, it will defer to the generic, which is the opposite of an AI engine’s purpose, as the goal is to reduce generic recommendations and become more and more specific. During their time as researchers at The University of Antwerp (where they also collaborated with researchers from Netflix), Froomle worked on this exact problem. They specifically looked at the confusion that arises in an AI engine when there are multiple, differing points of reference in a single profile, and developed an engine that recognizes distinct viewing, reading, and buying behavior. Instead of the engine trying to combine multiple interests into a small group of recommendations, the Froomle engine recognizes distinct patterns within a consumer’s behavior, isolates the interests that align with one another and makes recommendations within the parameters of the multiple patterns. For example, your children, even though they have their own Netflix profiles still use your profile to watch a vast array of happy, talking animated animals. You, on the other hand, enjoy horror flicks. The Froomle engine recognizes two distinct viewing behaviors, and instead of searching for a mash-up, it makes separate recommendations for each interest.

Ready to make your users your biggest fans?

Schedule a call to learn more on how we can help you make that happen