Personalizer

OVERVIEW

Create optimized user experiences

Boost conversion and engagement, and add real-time relevance to product recommendations, with reinforcement learning–based capabilities available only through Azure. Select hero content, optimize layouts, and personalize offers with two API calls. Use Personalizer, part of Azure AI Services, as a standalone personalization solution or to complement existing ranking engines—with no machine learning expertise required.

Give customers relevant experiences that improve over time based on their behavior. Improve click-through on home pages, create personalized channels, or optimize coupon offers and terms. Let AI discover what maximizes results to stay on top of changing trends. Unlike recommendation engines that offer a few options from a large catalog, Personalizer presents the single best outcome for a user every time they interact with your app.
Embed Personalizer by adding two lines of code. Easily inspect the accuracy of predictions and optimize as needed. Personalizer works with your data in any form. Start with no data or tap into an existing dataset to jump-start reinforcement learning. After it’s embedded, Personalizer can skip the learning curve. When you turn on apprentice mode, the service learns alongside your existing solution without being exposed to users until it meets your performance threshold.

Configure apprentice mode
Interpret and evaluate Personalizer through the interface. Validate results by reproducing them, and gauge effectiveness of parameters with a stack-ranked view. For example, evaluate the impact of age, gender, and time of day on content engagement.

Evaluate your models

FEATURES

Validate, monitor, and improve personalization with ease

Person participating in a video conference on a dual-monitor desk setup.

Pricing

Take charge with flexible pricing

Pay only for what you use, with no upfront costs. With Personalizer, you pay as you go based on number of transactions.

See Personalizer pricing

RESOURCES

Personalizer resources

Professional woman reading documents at a desk in an office.

Get started

Documentation

Learn more

Woman closely reviewing code and data on monitors.

Explore best practices

Responsible implementation guidelines

Learn more

See availability by region.
Personalizer and other AI Services offerings guarantee 99.9-percent availability. No SLA is provided for the Free pricing tier. See the SLA details.
Personalizer learns from the collective behavior of users, through the features or attributes about them that you send to the service. It uses this information to train a shared model that’s updated with information from every interaction, improving personalization outcomes for all users.
Personalizer works best when a rank call has 50 or fewer items. To personalize a choice from a larger list or catalog, reduce the number of items by using a recommendation engine or sorting technique. Learn more about scalability and performance.
Our algorithms and innovation are based on continuous work by Microsoft Research and on applied use of Personalizer in Microsoft businesses such as Xbox and Windows. Learn more about our approaches and reinforcement learning.
Personalizer works with or without user sign-in but delivers richer, more relevant interactions for signed-in users.
Yes. Personalizer learns in the background, in real time, via apprentice mode. When the service has learned enough from users’ actions to approximate your existing personalization system, you can use Personalizer in production with confidence to let it learn by interacting with your users and support a positive user experience.