The Query-Based Recommendations algorithm dynamically generates personalized suggestions by interpreting a shopper’s explicit search input or business-defined query. Ideal for headless environments, QBA transforms any keyword or rule into a tailored recommendation set—without relying on a graphical UI.

​
How It Works

  1. Query Parsing
    The system ingests the user’s search term or custom query, extracting keywords, filters, and logical operators.
  2. Intent & Preference Analysis
    Natural language processing and business rules interpret shopper intent (e.g., “leather office chair under $200”).
  3. Data Retrieval
    A tailored database lookup applies the parsed conditions—matching product attributes, categories, or metadata.
  4. Relevance Scoring
    Matched items are ranked by semantic relevance, performance signals (clicks, conversions), and any active merchandising rules.
  5. Results Delivery
    The algorithm returns a prioritized list of recommendations that best align with the original query’s intent and any configured rules.

​
Supported Rule Types

  • Global
  • Home Page
  • Product
  • Category
  • Cart
  • Search

​
When to Use

  • Headless Integrations: Embed personalized recommendations directly via API calls in any frontend (mobile app, CMS, PWA).
  • Rule-Driven Campaigns: Combine with business logic (“on sale,” “clearance”) for promotional or seasonal collections.
  • Zero-UI Experiences: Power chatbots or voice assistants where visual widgets aren’t available.

​
Example

  1. A shopper enters the query “ergonomic mesh office chair with lumbar support” into your site’s search bar.
  2. QBA parses the keywords and applies a rule to exclude out-of-stock items.
  3. It retrieves all matching chairs, ranks them by recent performance and in-stock levels, and applies any boost rules (e.g., promoted brands).
  4. The top 5 chairs are returned as recommendations in a headless widget on the category page.