The Golden Record measures how complete a product page’s data is for AI commerce platforms. It’s the percentage of product attributes correctly filled and structured out of the 34 that AI agents expect to recommend a product.
The concept originates from the Google Shopping Graph: stores with product data completeness at 99.9% see their AI visibility multiply 3-4x compared to stores with partial data.
Why data completeness changes everything
AI agents (ChatGPT Shopping, Perplexity Buy, Google AI Mode) don’t browse like humans. They extract structured data - price, availability, reviews, shipping - and compare across multiple products in milliseconds.
A product with missing data gets disqualified from the comparison. The AI cannot recommend a product when it doesn’t know the shipping cost, return policy, or customer reviews. It picks a competitor with complete data instead.
The 4 completeness tiers
| Tier | Threshold | AI Visibility | Meaning |
|---|---|---|---|
| Golden | ≥ 95% | 3-4x | Complete data, product is recommendable by all AI agents |
| Strong | 80–94% | 2x | Solid data, product appears in most recommendations |
| Moderate | 60–79% | 1x | Basic data, standard AI visibility level |
| Weak | < 60% | Ignored | Insufficient data, AI agents ignore the product |
The 7 categories and 34 attributes
1. Product Identity (7 attributes)
The foundational data that lets AI uniquely identify your product.
| Attribute | Weight | Description |
|---|---|---|
| Product title | 3 | Product name in schema.org name or extracted from page |
| Rich description | 3 | Description over 50 characters, structured and factual |
| Typed brand | 2 | Brand typed as Organization or Brand in schema.org |
| SKU | 1 | Unique stock identifier |
| GTIN/EAN | 2 | International barcode - critical for ChatGPT Shopping |
| MPN | 1 | Manufacturer Part Number reference |
| Product category | 1 | Category in schema.org or navigation |
2. Commerce (8 attributes)
The transactional data that lets AI agents compare offers.
| Attribute | Weight | Description |
|---|---|---|
| Structured price | 3 | Price in schema.org Offer.price |
| Currency | 1 | priceCurrency in schema.org |
| Availability | 2 | Stock via Offer.availability |
| Structured shipping | 2 | OfferShippingDetails with cost and delay |
| Shipping delay | 1.5 | deliveryTime or transitTime extractable |
| Return policy | 2 | MerchantReturnPolicy in schema.org |
| Return days | 1 | Extractable number of return days |
| Structured variants | 1.5 | ProductGroup or multiple Offer entries |
3. Trust & Social Proof (3 attributes)
The signals that let AI assess product reliability.
| Attribute | Weight | Description |
|---|---|---|
| Average rating | 2 | AggregateRating.ratingValue |
| Review count | 2 | AggregateRating.reviewCount > 0 |
| bestRating | 0.5 | Rating scale (e.g., /5) in schema.org |
4. Media (3 attributes)
The visuals that feed AI product carousels.
| Attribute | Weight | Description |
|---|---|---|
| Main image | 2 | image in schema.org Product |
| Multiple images (≥ 3) | 1 | At least 3 images for visual comparisons |
| Image alt text | 1 | At least one image with descriptive alt attribute |
5. Enriched Attributes (5 attributes)
Complementary data enabling fine-grained product comparisons.
| Attribute | Weight | Description |
|---|---|---|
| Structured attributes (≥ 3) | 1.5 | additionalProperty in schema.org |
| Material/composition | 1 | material in schema.org |
| Weight | 0.5 | weight in schema.org |
| Color | 0.5 | color in schema.org |
| Country of origin | 0.5 | countryOfOrigin in schema.org |
6. AI Discoverability (4 attributes)
Technical signals that help AI crawlers find and index your product.
| Attribute | Weight | Description |
|---|---|---|
| Complete Open Graph | 1 | og:image and og:title present |
| Canonical URL | 1 | canonical or url in schema.org |
| Product FAQ | 1 | FAQPage schema.org on product page |
| BreadcrumbList | 0.5 | Structured breadcrumb navigation |
7. AI Protocols (4 attributes)
Agentic commerce protocols that enable agents to interact with your store.
| Attribute | Weight | Description |
|---|---|---|
| llms.txt | 1.5 | LLM guidance file |
| UCP manifest | 1.5 | Universal Commerce Protocol (Google/Shopify) |
| MCP endpoint | 1.5 | Model Context Protocol (Anthropic) |
| Commerce bots allowed | 2 | GPTBot and PerplexityBot not blocked in robots.txt |
How to improve your Golden Record
Quick wins (high impact, low effort)
- Add GTIN/EAN in Shopify metafields - critical for ChatGPT Shopping
- Enable structured shipping via
OfferShippingDetailsin schema.org - Expose reviews in HTML (not JavaScript-only) so crawlers can read them
- Unblock AI crawlers in robots.txt (GPTBot, PerplexityBot)
Structural wins (high impact, medium effort)
- Add
MerchantReturnPolicywith return delay in days - Type your brand as
Organizationrather than plain text - Create a product FAQ with
FAQPageschema - Enrich
additionalPropertywith material, weight, color, country of origin
Advanced wins (medium impact, high effort)
- Deploy a UCP manifest (Universal Commerce Protocol)
- Create an llms.txt guiding AI to key content
- Expose an MCP endpoint for direct agent interactions
Golden Record vs GEO Score
| Aspect | Golden Record | GEO Score |
|---|---|---|
| Measures | Product data completeness | Overall AI visibility |
| Scope | 34 attributes of a product page | 9 factors across the entire site |
| Includes | Schema.org, media, protocols | Crawlers, content, trust, schema |
| Score | Percentage (0–100%) | Score out of 100 |
| Goal | Recommendable product page | Citable and recommendable site |
| Complementary | Good Golden Record without GEO Score = complete product but inaccessible site | Good GEO Score without Golden Record = accessible site but incomplete product pages |
How it maps to AI platforms
The Golden Record directly feeds compatibility with the 3 major platforms:
- Google AI Mode: the Shopping Graph uses completeness as its primary ranking factor
- ChatGPT Shopping: OpenAI’s product feed requires title, description, price, image, brand, GTIN - the 7 “required” Golden Record fields
- Perplexity Buy: product cards in answers are generated from structured data - the more complete the Golden Record, the richer the card
Frequently Asked Questions
What is the Golden Record in AI commerce?
The Golden Record is a product data completeness score. It measures 34 attributes across 7 categories (identity, commerce, trust, media, enrichment, discoverability, protocols). Stores reaching 95%+ completeness see 3-4x higher AI visibility based on Google Shopping Graph data.
How is the Golden Record Score calculated?
Each attribute has a weight (1 to 3 based on its importance for AI agents). The score is the ratio of present weights to total possible weights, converted to a percentage. Critical fields like title, description, and price weigh more than optional fields like weight or color.
What's the minimum threshold for AI visibility?
Stores below 60% completeness are generally ignored by AI agents. The baseline visibility threshold is 60%, but the optimal zone starts at 80% (2x visibility) and Golden Record status is achieved at 95%+ (3-4x visibility).
How can I quickly improve my Golden Record?
Focus on high-weight attributes still missing: GTIN/EAN, structured shipping, return policy, AggregateRating. These 4 attributes often represent the fastest wins for a Shopify store.
Does the Golden Record replace the GEO Score?
No. The GEO Score measures overall visibility (crawlers, content, trust signals). The Golden Record focuses exclusively on product page data completeness. Both are complementary: a good GEO Score without Golden Record means your site is accessible but your product pages lack data.