Product Recommendations

Boost sales with intelligent product recommendations powered by OpenSearch. The recommendation engine suggests relevant products to customers based on their browsing behavior and purchase history.

Overview

Navigate to Recommendations from the dashboard sidebar to configure and monitor product recommendations on your storefront.

The recommendation engine uses interaction data (views, cart additions, purchases, likes, and shares) to suggest relevant products to your customers.

Recommendation Settings

Enable/Disable

Toggle the entire recommendation system on or off. When disabled, no recommendation sections will appear on your storefront.

Recommendation Types

Enable or disable individual recommendation strategies:

Type	Description
Similar Products	Products similar to the one being viewed, based on category, tags, and interaction patterns
Trending Products	Products with the most views and purchases across your store
Category-Based	Products from the same category as the currently viewed product
Personalized	Recommendations tailored to the individual customer's browsing and purchase history

Display Locations

Control where recommendations appear on your storefront:

• Product Page — Show recommendations below the product details (enabled by default)
• Home Page — Show trending/personalized recommendations on the homepage
• Cart Page — Show complementary product suggestions on the cart page

Configuration

• Max Recommendations — Maximum number of products to show in each recommendation section (default: 8)
• Similarity Threshold — Minimum similarity score (0 to 1) for products to be considered "similar" (default: 0.3). Lower values show more results but may be less relevant.

Interaction Weights

Customize how different customer actions influence recommendations:

Action	Default Weight	Description
View	1	Customer viewed a product page
Like	2	Customer liked/favorited a product
Cart	3	Customer added product to cart
Share	4	Customer shared the product
Purchase	5	Customer purchased the product

Higher weights mean that action has more influence on which products get recommended. For example, with default weights, a purchase is 5x more influential than a view.

Recommendation Stats

The Stats section provides insights into how recommendations are performing:

• Total Interactions — Number of recommendation interactions tracked
• View Count — Total product views from recommendations
• Purchase Count — Products purchased after being recommended
• Interaction Breakdown — Distribution of interactions by type (view, cart, purchase, like, share)

Use these stats to monitor whether recommendations are driving engagement and conversions.

How It Works

1. Data Collection — As customers browse your storefront, their interactions (views, cart adds, purchases) are tracked and indexed in OpenSearch with product_view_count and product_purchase_count fields

2. Recommendation Generation — When a customer views a product, the engine queries OpenSearch to find similar products based on:

   • Product category and tags
   • Interaction patterns (products frequently viewed/purchased together)
   • Overall product popularity (trending)
   • Customer's personal history (personalized)

3. Display — Recommended products are displayed in the configured locations on the storefront

4. Continuous Learning — The engine improves over time as more interaction data is collected

Best Practices

• Start with Similar Products and Trending — These work well even with limited data
• Enable Personalized recommendations once you have enough customer interaction data
• Monitor the stats — If recommendation clicks are low, try adjusting the similarity threshold
• Keep max recommendations at 6-8 — Too many options can overwhelm customers
• Use Cart Page recommendations to increase average order value with complementary products