​
Glood.AI’s Recommendation Engine: AI Meets Merchant Control

Glood.AI’s recommendation platform combines cutting-edge artificial intelligence with intuitive merchant controls, giving you the best of both worlds—automated intelligence and business control. Our engine processes millions of data points in real-time while respecting your merchandising strategies configured in the Glood Dashboard.

​
Our AI Model Stack

​
Primary Models

  1. OpenAI GPT-4
    • Natural language understanding for product descriptions
    • Semantic similarity matching
    • Context-aware recommendations
    • Cross-lingual product matching
  2. Google Gemini Pro
    • Visual similarity analysis
    • Multi-modal product understanding
    • Style and aesthetic matching
    • Image-based recommendations
  3. Glood Proprietary Models
    • Fine-tuned on 100M+ e-commerce transactions
    • Optimized for conversion prediction
    • Real-time collaborative filtering
    • Sequential pattern mining

​
Recommendation Types Overview

Personalized For You

Individual-level personalization using deep learning

Frequently Bought Together

Association rule mining and purchase pattern analysis

Similar Products

Content-based and visual similarity matching

Trending Products

Real-time trend detection and momentum scoring

​
Glood Dashboard Configuration

Before diving into the AI, let’s understand how you control recommendations through the Glood Dashboard:

​
Section Configuration

  • Create Sections: Build recommendation sections with point-and-click interface
  • Choose Algorithm: Select from 13+ recommendation types
  • Set Parameters: Configure number of products, filtering rules, and display options
  • Preview Mode: Test recommendations before deploying to your store

​
Business Rules Setup

  • Product Exclusions: Block specific products or collections
  • Price Boundaries: Set min/max price ranges for recommendations
  • Inventory Thresholds: Hide products below certain stock levels
  • Brand Controls: Prioritize or exclude specific brands

​
How Glood Selects the Right Model

The Glood engine automatically selects the optimal AI model based on your configuration:

​
Data Signals We Analyze

​
User Signals

  • Browsing history and session data
  • Purchase history and frequency
  • Cart additions and abandonments
  • Time spent on products
  • Device and location context

​
Product Signals

  • Category and subcategory relationships
  • Price points and discount patterns
  • Inventory levels and availability
  • Product attributes and specifications
  • Visual features and style elements

​
Behavioral Signals

  • Click-through rates
  • Conversion rates by segment
  • Seasonal patterns
  • Cross-category affinities
  • Return and review data

​
Model Training & Optimization

​
Continuous Learning Pipeline

  1. Data Collection
    • Real-time event streaming from Shopify stores
    • Aggregated behavioral patterns
    • Purchase outcome tracking
  2. Feature Engineering
    • Automated feature extraction
    • Temporal pattern encoding
    • Cross-feature interactions
  3. Model Training
    • Daily retraining cycles
    • A/B testing new algorithms
    • Performance validation
  4. Deployment
    • Gradual rollout of improvements
    • Real-time monitoring
    • Automatic fallback mechanisms

​
Performance Optimization

​
Speed & Scalability

  • Caching Strategy: Multi-level caching for instant responses
  • Edge Computing: Recommendations served from global CDN
  • Batch Processing: Pre-computed recommendations for common scenarios
  • Real-time Inference: On-demand calculation for personalized results

​
Quality Metrics

We continuously monitor and optimize for:
  • Relevance Score: How well recommendations match user intent
  • Diversity Index: Variety in recommended products
  • Novelty Factor: Balance between familiar and new discoveries
  • Business Impact: Revenue lift and conversion improvement

​
Advanced Techniques

​
Ensemble Methods

Combining multiple models for superior results:
  • Weighted voting from different algorithms
  • Contextual bandits for exploration vs exploitation
  • Meta-learning for model selection

​
Deep Learning Architectures

  • Transformer models for sequence prediction
  • Graph neural networks for relationship modeling
  • Attention mechanisms for feature importance

​
Reinforcement Learning

  • Multi-armed bandits for recommendation optimization
  • Contextual bandits for personalization
  • Deep Q-learning for long-term value optimization

​
Glood Platform Features

​
Analytics Dashboard

Monitor your recommendation performance in real-time:
  • Conversion Tracking: See which recommendations drive sales
  • CTR Analysis: Click-through rates by section and algorithm
  • Revenue Attribution: Direct revenue impact of each recommendation type
  • A/B Testing: Built-in testing framework for optimization

​
API Access

Integrate Glood recommendations anywhere: 
GET https://api.glood.ai/v1/recommendations
{
  "shop": "your-store.myshopify.com",
  "type": "similar_products",
  "product_id": "7234567890",
  "limit": 4,
  "filters": {
    "in_stock": true,
    "price_range": [10, 100]
  }
}

​
Customization Options

Every aspect of Glood recommendations can be customized:
  • Visual Templates: Match your store’s design perfectly
  • Responsive Layouts: Optimized for mobile and desktop
  • Language Support: Multi-language recommendations
  • Currency Handling: Automatic currency conversion

​
Explore Glood’s Recommendation Algorithms

Dive deeper into how each Glood recommendation type works: