In the growingly competitive world of online stores, delivering relevant product recommendations is an important part of customer experience. If customers see truly relevant related products on the product page, it can actually lead to increased sales. Text embedding similarity search, powered by AI models and integrated with Magento through powerful search engines like Elasticsearch (since version 8) or OpenSearch (since version 2), offers real measurable benefits for product recommendation.
From Manual Work to Smart Automation
Imagine you have a store with thousands of products. Now think about how much work it takes to open each product in your platform’s admin interface, decide on suitable related, cross-sell, and upsell products, search for them, manually select the right ones, save the changes. Then do it all over again for every single product. This amount of work could take weeks!
Now imagine doing all of that with just a couple of clicks. Sounds good, right?
Relevant Recommendations with the Help of AI Models

AI embedding models turn product text into vectors – basically numerical representations of each product.
Magento uses these vectors to perform similarity searches through Elasticsearch or OpenSearch. This allows it to identify and recommend products that are most similar to the one a customer is viewing.
For example, if the product is a window, it might suggest curtains or glass cleaners. If it’s shoes, it could recommend extra laces or a shoe cleaning kit. This eventually leads to highly contextually relevant product recommendations that increase conversion rates and improve customer satisfaction.
Text Embedding Similarity Search
Text is the most consistent and scalable source of product metadata in eCommerce. AI text embedding models convert textual data such as product titles, description and other attributes into dense numerical vectors (embeddings) that capture semantic meaning beyond simple keywords. These embeddings allow Elasticsearch or Opensearch to measure similarity between texts based on meaning rather than exact word matches, enabling more relevant product recommendation capabilities.
Key Benefits of Text Embeddings in eCommerce Product Recommendation
This technology brings several practical benefits:
1. Improved Understanding of Product Context
Embedding similarity search “understands” the intent behind queries and product descriptions. Semantic matching increases the relevance of matching products.
2. Scalability for Large Catalogs
Embedding-based search scales efficiently to millions of products by precomputing embeddings during indexing and using vector databases for fast similarity comparisons via cosine similarity. This is especially valuable for large Magento catalogs, where manual intervention could take a huge amount of time to handle the entire catalog.
3. Ambiguity, Typos, and Multilingual Content
Text embeddings handle ambiguous queries (like “apple” as fruit or a brand) by considering the entire context. They also tolerate typos and language variations by focusing on semantic similarity rather than exact text matches. This flexibility improves the relevance of product recommendations, whatever the context of descriptions or attributes is.
4. Hybrid Search
Combining embedding similarity with traditional keyword search in a hybrid model allows eCommerce platforms to leverage the precision of lexical search and the flexibility of semantic search. This ensures that specific products get exact matches, while broader or universal terms benefit from semantic understanding.
How It All Comes Together: Magento, AI, and Vector Search
- Elasticsearch and Opensearch support vector search capabilities, enabling efficient storage and retrieval of product embeddings using similarity metrics.
- Text Embedding AI Models (Sentence Transformers, provided by most AI tools today like OpenAI and Google or open source models available on Huggingface) encode product data and user queries into semantic vectors.
- Magento serves as the eCommerce platform where product catalogs are managed and related, upsell and cross-sell product recommendations are displayed, using embedding similarity search results to enhance the related products feature.
- By integrating these technologies, eCommerce sites can deliver fast and accurate product recommendations that improve user engagement and drive sales.
Text embedding similarity search in eCommerce, powered by Elasticsearch or Opensearch and AI models, revolutionizes product recommendations by understanding semantic meaning, giving more accurate suggestions and scaling effortlessly. When combined with Magento, it creates a robust system that gives the customers more relevant recommendations and, if done right, it could actually boost sales!
Potential, Challenges, Solution
Building on the concepts explored above, we recognized the tremendous potential to transform how eCommerce stores handle product recommendations.
However, while the technology offers incredible accuracy and scalability, implementing it in a real-world store environment posed several challenges. These include handling large, diverse catalogs, managing different recommendation types (related, upsell, cross-sell), and ensuring the system remains easy for merchants to configure and maintain.
To bridge this gap, we developed our AI-Powered Recommendation Engine: a comprehensive, automated solution that combines the power of AI semantic similarity with smart business logic and data-driven insights.
AI-Powered Product Recommendation Engine
The AI-Powered Recommendation Engine is our answer to the common pain points in product recommendation. This all-in-one product recommendation engine is designed to help store owners increase sales, improve customer experience, and streamline product discovery by eliminating manual catalog work.
This solution simplifies the entire recommendation process by dynamically generating relevant suggestions using text embeddings, sales data, and configurable rules. Merchants no longer need to spend weeks manually editing related items, as the engine adapts to their catalog, updates recommendations in real time, and helps drive higher conversions and customer satisfaction.
How Does It Work?
This powerful new functionality combines three advanced components to automatically suggest related, up-sell, cross-sell and frequently bought together products on your storefront:
Automated Related Products
Leverages a smart set of business rules to dynamically fill related product slots on the product page. It prioritizes manually selected products, then pulls from a chosen category, the product’s deepest categories, or best-sellers from the last month – ensuring your related products are always relevant and fresh.

Frequently Bought Together (FBT)
Analyzes historical sales data using a Frequent Pattern Growth algorithm to identify products customers commonly purchase together. This data-driven approach automatically generates product sets to encourage cross-selling opportunities. The FBT data is kept up to date with scheduled or manual reindexing, configurable via the admin panel or CLI commands.

AI Recommendation Engine
Harnesses the power of AI and vector search technologies (OpenAI, Google Gemini, or LocalAI combined with Elasticsearch or OpenSearch) to generate highly personalized related, up-sell, and cross-sell product recommendations. It converts product titles, descriptions, and categories into vector embeddings and finds similar products based on configurable similarity metrics and thresholds.

Key Features & Benefits
- Flexible Configuration: Each of the three components can be enabled individually or together, tailored to your catalog and marketing goals.
- Priority Handling: Manual selections always take precedence, with fallback logic ensuring all recommendation slots are filled intelligently.
- Data-Driven Recommendations: Sales data and advanced algorithms ensure customers see what they really want, boosting conversions.
- AI-Powered: Semantic similarity analysis delivers smarter, contextually relevant recommendations that evolve with your catalog.
- Admin-Friendly: Easily configure max product counts, categories, frequency of updates, AI providers, and similarity settings – all from your admin dashboard.
- Seamless Storefront Integration: Recommendations appear naturally alongside manually added products, without confusing labels or UI clutter.
Why This Matters
In an era where online shopping is more crowded than ever, personalized product suggestions are a proven way to increase average order value and improve customer satisfaction. Our Recommendation Engine automates this process, saving countless catalog editing hours while delivering optimized, real-time recommendations backed by AI and various backend data analysis techniques we developed.
Whether you want a simple related products solution or an advanced AI-driven recommendation system, this new functionality provides the tools you need!
Hootify + AI Recommendations: A Smarter eCommerce Experience

We’re excited to share that the Recommendation Engine is now integrated into Hootify, our Magento-based eCommerce solution. With Hootify, you get this powerful tool from day one: fully embedded into the admin, easy to configure, and ready to scale.
Learn more about our Recommendation Engine by exploring the Hootify User Guide for technical insights. Or request a free admin demo to see how it works from a merchant’s perspective.
Get Started Today
Backed by years of experience building and optimizing Magento stores, we developed this AI-Powered Recommendation Engine to address the very real pain points faced by merchants. If you’re running a Magento store and want smarter, more relevant product recommendations without the manual effort, this solution was built for you.
Contact us and let’s explore the difference AI and automation can make for your store!