What is Personalizer?
Azure Personalizer is a cloud-based service that helps your applications choose the best content item to show your users. You can use the Personalizer service to determine what product to suggest to shoppers or to figure out the optimal position for an advertisement. After the content is shown to the user, your application monitors the user's reaction and reports a reward score back to the Personalizer service. This ensures continuous improvement of the machine learning model, and Personalizer's ability to select the best content item based on the contextual information it receives.
Content is any unit of information, such as text, images, URL, emails, or anything else that you want to select from and show to your users.
This documentation contains the following article types:
- Quickstarts are getting-started instructions to guide you through making requests to the service.
- How-to guides contain instructions for using the service in more specific or customized ways.
- Concepts provide in-depth explanations of the service functionality and features.
- Tutorials are longer guides that show you how to use the service as a component in broader business solutions.
Before you get started, try out Personalizer with this interactive demo.
How does Personalizer select the best content item?
Personalizer uses reinforcement learning to select the best item (action) based on collective behavior and reward scores across all users. Actions are the content items, such as news articles, specific movies, or products.
The Rank call takes the action item, along with features of the action, and context features to select the top action item:
- Actions with features - content items with features specific to each item
- Context features - features of your users, their context or their environment when using your app
The Rank call returns the ID of which content item, action, to show to the user, in the Reward Action ID field.
The action shown to the user is chosen with machine learning models, that try to maximize the total amount of rewards over time.
Let's take a look at a few scenarios where Personalizer can be used to select the best content to render for a user.
|Content type||Actions (with features)||Context features||Returned Reward Action ID
(display this content)
|Device news is read from
Month, or season
|Device movie is watched from
Type of user
|Device shopping is read from
Spending tier of user
Month, or season
Personalizer used reinforcement learning to select the single best action, known as reward action ID. The machine learning model uses:
- A trained model - information previously received from the personalize service used to improve the machine learning model
- Current data - specific actions with features and context features
When to use Personalizer
Personalizer's Rank API is called each time your application presents content. This is known as an event, noted with an event ID.
Personalizer's Reward API can be called in real-time or delayed to better fit your infrastructure. You determine the reward score based on your business needs. The reward score is between 0 and 1. That can be a single value such as 1 for good, and 0 for bad, or a number produced by an algorithm you create considering your business goals and metrics.
Use Personalizer when your content:
- Has a limited set of actions or items (max of ~50) to select from in each personalization event. If you have a larger list, use a recommendation engine to reduce the list down to 50 items for each time you call Rank on the Personalizer service.
- Has information describing the content you want ranked: actions with features and context features.
- Has a minimum of ~1k/day content-related events for Personalizer to be effective. If Personalizer doesn't receive the minimum traffic required, the service takes longer to determine the single best content item.
Since Personalizer uses collective information in near real-time to return the single best content item, the service doesn't:
- Persist and manage user profile information
- Log individual users' preferences or history
- Require cleaned and labeled content
How to design for and implement Personalizer
Design and plan for content, actions, and context. Determine the reward algorithm for the reward score.
Each Personalizer Resource you create is considered one Learning Loop. The loop will receive the both the Rank and Reward calls for that content or user experience.
Resource type Purpose Apprentice mode
Train the Personalizer model without impacting your existing application, then deploy to Online learning behavior to a production environment Standard,
Online learning behavior in a production environment Free,
Try Online learning behavior in a non-production environment
Add Personalizer to your application, website, or system:
- Add a Rank call to Personalizer in your application, website, or system to determine best, single content item before the content is shown to the user.
- Display best, single content item, which is the returned reward action ID, to user.
- Apply business logic to collected information about how the user behaved, to determine the reward score, such as:
Behavior Calculated reward score User selected best, single content item (reward action ID) 1 User selected other content 0 User paused, scrolling around indecisively, before selecting best, single content item (reward action ID) 0.5
- Add a Reward call sending a reward score between 0 and 1
- Immediately after showing your content
- Or sometime later in an offline system
- Evaluate your loop with an offline evaluation after a period of use. An offline evaluation allows you to test and assess the effectiveness of the Personalizer Service without changing your code or affecting user experience.
- Personalizer C#/.NET SDK
- Personalizer Go SDK
- Personalizer Python SDK
- REST APIs