Overview - Advanced Recommendations

Advanced recommendations use machine learning algorithms and neural networks to discover accurate recommendations. These recommendations are dynamic and adapt to user interaction patterns over time and update almost in real-time.

The recommendation results are sorted in relevant order which means the first recommended item is most relevant. These recommendations can be further enlisted in the following categories:

Recommendation models

Recommended Items

Just for you! Tailored recommendations for each user, derived from individual interaction behaviors and influenced by wider user trends. Think of this as a personal stylist who knows you better than your best friend. They curate choices based on your interactions and weave in preferences from those who engage just like you.

Model Ingredients:
User Action events: Product Viewed, Add to Wishlist, Add to Cart, Product Purchased

Example Use Cases

  • Provide users with the next best items.
  • Formulate the 'You May Like', 'You May Be Interested', or 'Recommended For You' collections.

Screenshot 2024-02-05 at 3.22.37 AM.png

Similar Items

Ever eyed an item and thought, 'I love this, but is there more like it?' Meet our Similar Item model. It digs through shared item attributes and past trends to find items that match the charm of your favorites.

Model Ingredients:
User Action events: Product Viewed, Add to Wishlist, Add to Cart, Product Purchased

Catalog Attributes: Title, Price, Category, Brand (but customizable as per your need)

Example Use Cases

  • Provide users with alternative options for products or services they are interested in.
  • Help users explore a wider range of products or services that match their interests.

Screenshot 2024-02-05 at 3.22.49 AM.png

Frequently Viewed Together Items

It's like the detective of recommendations. Sherpa AI finds hidden links between items, showing you what others often view in tandem, and making cross-selling a breeze. we present cross-selling opportunities you didn't know existed.

Model Ingredients:
User Action events: Product Viewed

Example Use Cases

  • Provide users with relevant product bundles or packages.
  • Increase sales and engagement by suggesting additional related products or services.

Screenshot 2024-02-05 at 3.22.58 AM.png

Frequently Bought Together Items

Enhance the shopping experience by suggesting items that are often bought together, paving the way for upselling and bundled convenience. This model is more of a savvy shopkeeper who always knows the perfect bundle of items that usually find their way to the checkout counter together.

Model Ingredients:
User Action events: Product Purchased

Example Use Cases

  • Suggest complementary products or services to increase sales and engagement.
  • Improve customer satisfaction by anticipating their needs and providing relevant suggestions to complete their purchase.

Screenshot 2024-02-05 at 3.23.05 AM.png

Popular Items

The crowd-pleasers. When in doubt, turn to what’s trending. Find out which items are creating the buzz. These social butterfly recommendations are defined as default fallback. For the newbies with no prior interactions, these items will be your welcoming committee, ensuring your new buyers are always in the loop.

Model Ingredients:
User Action events: Product Viewed, Add to Wishlist, Add to Cart, Product Purchased

info

Note

These recommendations are stitched as the default fallback for all the Advanced recommendations models and are served if a user has not engaged with any items in the last 2 months.

 

Example Use Cases

  • Activate new users by exhibiting trends in your platform.
  • Help discover new and popular products or services that they may be interested in.

Filters

The filter settings allow you to refine the recommendations shown to your customers. These filters work alongside the item attribute model rule to enhance the relevance of the suggested items.

You can filter items in two ways:

  • Keep Items: This filter allows you to define user action criteria that will include only those items that are common in both the results of item attribute rules and user action filter criteria.
  • Remove Items: This filter allows you to define user action criteria that will remove items from the results of item attribute rules.
info

Information


The final product calculation after applying recommendation filters looks as follows:

  • A = products from the core recommendation filter
  • Keep items:
    • B = products from the 'keep item' filter setting
    • Final result = A B
  • Remove items:
    • C = products from the 'remove item' filter setting
  • Final result = A B - C

Example use cases

  • Recommend a frequently bought together bundle but remove the items added to the cart.
  • Keep only those Recommended items that the user has viewed in the last 30 days.
  • Remove the viewed items from the Frequently viewed together items list to encourage new product discovery.
  • Remove the items that the user has already added to wishlist from Similar items recommendations.

Previous

Next

Was this article helpful?
0 out of 0 found this helpful

How can we improve this article?