Skip to content

Item recommendations

Step up your recommendation game with Braze by creating a recommendation engine that can suggest items and content to your users that they actually want. From customizing experiences with AI to building your own engines with Liquid or Connected Content, you’ll find everything you need to make every recommendation count.

Prerequisites

Before you can create or use item recommendations in Braze, you’ll need to create at least one catalog—only items from that catalog will be recommended to users.

Types and use cases

AI Personalized

As part of the AI item recommendations feature, AI Personalized recommendations take advantage of deep learning to predict what your users are most likely to be interested in next based on what they’ve shown interest in in the past. This method provides a dynamic and tailored recommendation system that adapts to user behavior.

AI Personalized recommendations use the last 6 months of item interaction data, like purchases or custom events, to build the recommendation model. For users without enough data for a personalized list, the most popular items serve as a fallback so your users are still getting relevant suggestions.

With AI item recommendations, you can also further filter the items available with selections. However, selections with Liquid cannot be used in AI recommendations, so keep that in mind when building your catalog selections.

Use cases

Based on the interaction data being tracked, use cases for this model could include:

Predict and recommend the items a user is most likely to purchase next, based on purchase events or custom events related to purchases. For example:

  • A travel site could suggest vacation packages, flights, or hotel stays based on a user’s browsing history and previous bookings, anticipating their next travel destination and making it easier for them to plan their trip.
  • A streaming platform can analyze viewing habits to recommend shows or movies a user is most likely to watch next, keeping them engaged and reducing churn rates.
Requirements
  • AI item recommendations
  • Catalog of relevant items
  • A method for tracking purchases, either a purchase object or custom event
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to AI Personalized.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose how you currently track purchase events and the corresponding event property.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

The “Most popular” recommendation model features items that users engage with most.

Use cases

Based on the interaction data being tracked, use cases for this model could include recommending:

Encourage users to explore popular items in your catalog based on purchases. To make sure you’re only surfacing relevant content, we recommend filtering with a selection. For example, a food delivery service could highlight top-rated dishes or restaurants within a user’s area, based on the popularity of orders across the platform, encouraging trial and discovery.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • A purchase object or any custom event
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most popular.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items. For example, the food delivery service might have a selection to filter for restaurant location or type of dish.
  5. Choose how you currently track events and the corresponding event property.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Encourage users to explore items that they’ve recently liked or items that are popularly liked, based on a custom event for likes. For example, a music streaming app could create personalized playlists or suggest new album releases based on the genres or artists a user has liked in the past, enhancing user engagement and time spent on the app.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • Custom event for likes
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most recent.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose Custom Event and select your custom event for likes from the list.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Highlight items that have gained attention across your user base through views to encourage engagement or purchases. For example, a real estate website could display the most viewed listings in a user’s search area to highlight properties that are attracting a lot of attention, potentially indicating good deals or desirable locations.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • Custom event for views
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most popular.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose Custom Event and select your custom event for views from the list.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Showcase items that are added to carts by many other shoppers, providing users with a glimpse into the current trends among your offerings.

For example, a fashion retailer could promote clothes and accessories that are trending based on popular additions to carts by other customers. They can then create a dynamic “Trending Now” section on their homepage and mobile app, which updates in real-time to encourage shoppers to purchase before items sell out.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • Custom event for added to cart
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most popular.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose Custom Event and select your custom event for added to cart from the list.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Most recent item

The “Most recent” recommendation model features items that users engage with most. Use this model to reduce churn by encouraging lapsing users to re-engage with relevant content.

Use cases

Based on the interaction data being tracked, use cases for this model could include recommending:

Encourage users to revisit items that they’ve recently clicked on, based on a custom event for clicks. For example, an online fashion retailer could create a recommendation to send follow-up emails or push notifications featuring clothes that a user has shown interest in by clicking on them, encouraging the user to revisit the item and make a purchase.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • Custom event for clicks
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most recent.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose Custom Event and select your custom event for clicks from the list.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Encourage users to explore items that they’ve recently liked or items that are popularly liked, based on a custom event for likes. For example, a music streaming app could create personalized playlists or suggest new album releases based on the genres or artists a user has liked in the past, enhancing user engagement and time spent on the app.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • Custom event for likes
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most recent.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose Custom Event and select your custom event for likes from the list.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Promote items that users have recently interacted with, including views, clicks, or purchases. This approach keeps your recommendations fresh and aligned with the user’s latest interests​. For example:

  • Education: An online education platform could encourage users who have recently watched an educational video but haven’t enrolled in a course to check out similar courses or subjects of interest to keep the user engaged and motivated to start learning.
  • Fitness: A fitness app can suggest workouts or challenges that are similar to the ones a user has recently completed or interacted with, keeping their exercise routine varied and engaging.
  • Home improvement retailer: After a customer purchases a power tool, a home improvement retailer can recommend related accessories or safety gear based on their recent purchase, enhancing the user’s experience and safety.
Requirements
  • AI item recommendations
  • Catalog of relevant items
  • A purchase object or any custom event for an engagement interaction
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most recent.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose Custom Event and select your custom event for clicks from the list.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Remind users of their interest in items that they recently added to their cart, but haven’t purchased yet. For example, an online retailer could send reminders or offer limited-time discounts on the items in their cart, encouraging users to complete their purchases before the offers expire.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • Custom event for added to cart
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Most recent.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose Custom Event and select your custom event for added to cart from the list.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

The “Trending” recommendation model features items that had the most positive momentum when it comes to recent user interactions.

Unlike the “Most Popular” model, which features items with consistently high interaction, this model features items that have experienced an uptick in interactions. You can use it to recommend products that are up-and-coming, and currently seeing increased traction.

Use cases

Based on the interaction data being tracked, use cases for this model could include recommending:

Highlight items that your users have recently purchased with increased frequency. For example, an eCommerce business could recommend seasonal items that users are starting to stock up on during their preparations for the next season.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • A method for tracking purchases (either a purchase object or custom event)
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Trending.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose either a purchase event or a custom event that tracks purchases, along with the corresponding property.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Highlight items that your users have recently liked with increased frequency. For example, a music app could feature up-and-coming artists who have experienced a recent surge in user likes.

Requirements
  • AI item recommendations
  • Catalog of relevant items
  • Custom event for tracking likes
Setting it up
  1. Create an AI item recommendation.
  2. Set the Type to Trending.
  3. Select your catalog.
  4. (Optional) Add a selection to filter your recommendation to only relevant items.
  5. Choose your custom event for tracking likes, along with the corresponding property.
  6. Train the recommendation.
  7. Use the recommendation in messaging.

Selections-based

Selections are specific groups of catalog data. When you use a selection, you’re essentially setting up custom filters based on specific columns in your catalog. This could include filters for brand, size, location, date added, and more. It gives you control over what you’re recommending by allowing you to define criteria that items must meet to be shown to users.

The previous three types all involve setting up and training a recommendation model in Braze. While you can use selections in those models as well, you can also accomplish some recommendation use cases with just catalog selections and Liquid personalization.

Use cases

Based on the interaction data being tracked, use cases for this model could include recommending:

This scenario doesn’t rely directly on user actions but rather on catalog data. You can filter for new items based on their addition date to the catalog and promote these through targeted campaigns or Canvases without needing to train a recommendation model.

For example, a tech eCommerce platform could alert tech enthusiasts about the latest gadgets or upcoming pre-orders, using filters to target items that have been recently added to the catalog.

Requirements
  • Catalog of relevant items with a field for date added
Setting it up
  1. Create a selection based on your catalog. Make sure your catalog has a time field (field with a Data type set to Time) that corresponds to the date the item was added.
  2. (Optional) Add any filters if desired.
  3. Make sure Randomize Sort Order is turned off.
  4. For Sort Field, select your date added field.
  5. Set Sort Order to descending.
  6. Use the selection in messaging.

For a diverse user experience, recommending random items can introduce variety and potentially spark interest in less-visited catalog areas. This method doesn’t require specific models or events but rather uses a catalog selection to ensure items are displayed randomly.

For example, an online bookstore could offer a “Surprise Me” feature, recommending a random book based on the user’s past purchases or browsing habits, encouraging exploration outside of their normal reading genres.

Requirements
  • Catalog of relevant items
  • Selection with Randomize Sort Order turned on
Setting it up
  1. Create a selection based on your catalog.
  2. (Optional) Add any filters if desired.
  3. Turn on Randomize Sort Order.
  4. Use the selection in messaging.

Rules-based

A rules-based recommendation engine uses user data and product information to suggest relevant items to users within messages. It uses Liquid and either Braze catalogs or Connected Content to dynamically personalize content based on user behavior and attributes.

Rules-based recommendations are based on fixed logic that you must manually set. This means your recommendations won’t adjust to a user’s individual purchase history and tastes unless you update the logic, therefore this method is best for recommendations that don’t need frequent updates.

Use cases

Based on the interaction data being tracked, use cases for this model could include:

  • Restock reminders: Sending restock reminders for items with a predictable usage cycle, like monthly vitamins or weekly groceries, based on their last purchase date.
  • First-time buyers: Recommend starter kits or introductory offers to first-time buyers to encourage a second purchase. Loyalty programs: Highlight products that would maximize a customer’s loyalty points or rewards based on their current points balance.
  • Educational content: Suggest new courses or content based on the topics of previously consumed or purchased materials.
Github   Edit this page on GitHub
HOW HELPFUL WAS THIS PAGE?
New Stuff!