When it comes to recommendation engines, it's important to consider the specific needs and preferences of each user. While some models, like content-based recommendation systems, may be better suited for users with a limited history of engagement, more complex models may be better suited for users with a longer trail of feedback.

Content-based models rely on the attributes of the products themselves to make recommendations. This approach can be particularly effective for users with a limited history of engagement, as it doesn't rely on past behavior to make recommendations. Instead, it focuses on the specific characteristics of the products that a user has shown interest in.

On the other hand, more complex models that leverage data on past behavior can be more effective for users with a longer trail of feedback. These models can provide a more nuanced and personalized experience, as they take into account the unique interests and preferences of each user.

It's important to note that the effectiveness of recommendation engines can be influenced by a variety of factors, including the quality of the data being used and the specific needs of each user. In order to fully understand these nuances and ensure that recommendations are both accurate and effective, it's important to perform quality assurance (QA) testing on recommendation engines.

‍

User Cohort List

Create a User Cohort

You can create a user cohort currently in your dashboard in 3 differents way:

• Uplaoding your own list of user id that you would like to see considered as a user allowlist
• Generating the list by segmenting user based on theier count of ratings and engagement
• Leveraging Crossing Minds ML Team to generate tast based cohorts and special allowlist that could also be exported to your marketing platforms

Recommendations for a Cohort

To evaluate and receive user based recommendation for users belonging to a specific allowlist:

1. Open the User Allowlist section on the right side of your screen
2. Select a cohort on which you'd like to evaluate your recommendations
3. Click the "select random id" button to pick a user belonging to this allowlist