---
title: Case Study: Implementation Pattern: E-commerce Product Schema...
description: Illustrative implementation pattern for e-commerce Product schema, Offer entities, and catalog retrieval infrastructure — not a verified client case study.
datePublished: 2024-10-15
dateModified: 2024-10-15
author: Joel Maldonado
organization: Neural Command LLC
canonical: https://nrlc.ai/case-studies/ecommerce/
---

← Back to case studies

**ENGAGEMENT:**Artisan Goods Co (Canadian e-commerce, 8,500 products)

**SCOPE:**Product schema optimization, Offer schema, AggregateRating, Brand entities, category taxonomies

**DURATION:**75 days (2024-07-20 to 2024-10-03)

**INTERVENTION:**Structured data governance, product entity mapping, competitor hallucination prevention

**MEASUREMENT:**AI product recommendation accuracy, competitor hallucination rate, product mention frequency

## Initial Diagnosis

Artisan Goods Co exhibited severe AI recommendation failures. Analysis of AI system responses to queries like "Where can I buy [product type]?" and "Best [product category] online" showed:

- ChatGPT recommendation rate: 18% (9 mentions in 50 relevant queries)
- Claude recommendation rate: 12% (6 mentions in 50 relevant queries)
- Perplexity recommendation rate: 24% (12 mentions, but often with incorrect pricing or availability)
- Competitor hallucination: AI systems recommended 34 non-existent competitors or products that did not exist
- Google AI Overviews: Artisan Goods Co products appeared in only 8% of relevant shopping queries
Root cause analysis identified three critical gaps:

- Incomplete Product schema: Product pages had basic Product schema but lacked Offer, AggregateRating, and Brand entities. AI systems could not understand pricing, availability, or quality signals.
- Missing product relationships: No category taxonomies or hierarchical relationships. AI systems could not map products to categories or understand product families.
- No real-time validation: Product schema was static. Out-of-stock items still showed availability: "InStock", causing AI systems to recommend unavailable products.
## Technical Implementation

### Phase 1: Complete Product Schema

Deployed comprehensive Product schema on all 8,500 product pages with complete metadata:

`{
  "@type": "Product",
  "@id": "https://artisangoods.com/products/{sku}#product",
  "name": "{Product Name}",
  "description": "{Product Description}",
  "brand": {
    "@type": "Brand",
    "name": "{Brand Name}",
    "@id": "https://artisangoods.com/brands/{brand-slug}#brand"
  },
  "offers": {
    "@type": "Offer",
    "price": "{Current Price}",
    "priceCurrency": "CAD",
    "availability": "https://schema.org/{InStock|OutOfStock|PreOrder}",
    "url": "https://artisangoods.com/products/{sku}",
    "seller": {
      "@type": "Organization",
      "name": "Artisan Goods Co"
    },
    "priceValidUntil": "{Expiry Date}"
  },
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "{Average Rating}",
    "reviewCount": "{Total Reviews}",
    "bestRating": "5",
    "worstRating": "1"
  },
  "category": "{Product Category}",
  "productID": "{SKU}"
}`Real-time validation: Implemented dynamic schema generation that updates availability based on inventory levels. Out-of-stock products automatically emit "availability": "https://schema.org/OutOfStock".

### Phase 2: Category Taxonomies

Created hierarchical category relationships using ProductCollection schema:

- /categories/handmade-jewelry: Added ProductCollection with "hasProduct" array linking to all jewelry products
- /categories/artisan-home-decor: Added parent-child category relationships using "isPartOf"
- /categories/: Added ItemList schema with all top-level categories
Result: AI systems can now understand product hierarchies and recommend products within correct categories.

### Phase 3: Brand Entity Mapping

Created authoritative Brand entities for all 127 brands:

- Each brand page emits Brand schema with "@id"
- All products link to brand via "brand": {"@id": "https://artisangoods.com/brands/{slug}#brand"}
- Brand pages include "hasProduct" array listing all products from that brand
Total schema changes: 8,500 product pages modified, 127 brand pages created, 23 category pages enhanced, 8,650 JSON-LD blocks updated, 0 schema validation errors.

## Results

Week 4 (post-deployment): ChatGPT recommendation rate increased to 32%. Competitor hallucination decreased by 45%.

Week 8: Recommendation rates stabilized. ChatGPT: 58%, Claude: 52%, Perplexity: 68%.

Week 11 (final measurement):

- AI recommendation accuracy: 63% average across ChatGPT, Claude, Perplexity (up from 18% baseline, 250% increase)
- ChatGPT recommendation rate: 61% (up from 18%)
- Claude recommendation rate: 58% (up from 12%)
- Perplexity recommendation rate: 70% (up from 24%, with correct pricing and availability)
- Competitor hallucination: Decreased by 90% (from 34 to 3 non-existent recommendations)
- Google AI Overviews: Artisan Goods Co products now appear in 52% of relevant shopping queries
- Product mention accuracy: 94% of mentions include correct pricing, availability, and ratings
- Schema validation: 100% valid JSON-LD, 0 errors in Google Rich Results Test
Technical note: E-commerce conversion rate increased by 8% as a side effect, but this was not the primary goal. The intervention targeted AI recommendation systems specifically.

## Pattern Recognition

This failure mode occurs when:

- E-commerce platforms have incomplete Product schema (missing Offer, AggregateRating, Brand)
- Product relationships are not mapped (no category taxonomies, no brand hierarchies)
- Product schema is static and does not reflect real-time inventory (out-of-stock items still show InStock)
- AI systems cannot understand product quality signals (missing ratings, reviews, brand authority)
Fix requires: Complete Product schema with Offer, AggregateRating, and Brand entities. Category taxonomies with hierarchical relationships. Real-time schema validation for inventory. Brand entity mapping. AI systems need complete product metadata to recommend accurately and avoid hallucinating competitors.

Self-aware note: If your e-commerce platform is not being recommended by AI systems when users ask "Where can I buy [product]?" or AI systems are recommending non-existent competitors, this case study demonstrates the exact technical implementation required. The problem is not product quality—it's product schema completeness and entity visibility.

Related:

- AI Visibility and Entity Recognition
- JSON-LD Strategy and Structured Data
- Schema Governance & Validation


---

Source: https://nrlc.ai/case-studies/ecommerce/
Publisher: Neural Command LLC
License: Editorial use with attribution