HONG KONG JEWELLERY PRODUCT-DATA GUIDE · 5 NOV 2026

How can a Hong Kong jewellery website
help AI understand its products?

For a jewellery website, the first step towards clearer search and AI visibility is not publishing more “AI content”. It is giving every product a consistent, verifiable identity: material, gemstone origin, certificate details, price, availability and a clear route to purchase or enquire. AI can help organise approved information and answer routine questions. It should not guess whether a diamond is natural or laboratory-grown, whether a certificate belongs to a particular item, or whether today’s gold-linked price and stock status are still valid.

Author: CLEARCRAFT LAB editorial teamVersion: v1.0 · official sources checked

Summary: This article gives Hong Kong jewellery businesses a practical product-data and human-enquiry workflow, with the boundaries of Google Search, Product structured data and Merchant Center clearly separated. It is not a promise of ranking, citation or revenue from Google, Gemini or another AI platform.

Start with one verifiable record for each product

The same ring design may be offered in 18K white gold, yellow gold or platinum. It may also use a natural diamond, a laboratory-grown diamond or different centre stones. Similar images do not make these items interchangeable. Each product should have a stable product ID or SKU; where variants exist, each variant should also be uniquely identifiable. The page should clearly show the relevant name, material, stone details, size or weight, images, supply status and purchase method.

Where one design varies only by material, colour or size, the relationship can be modelled explicitly. Google’s Product variant documentation supports ProductGroup, hasVariant, variesBy and productGroupID, and calls for a unique identifier for each variant. This can help a system understand how products relate to each other; it is not an AI recommendation switch.

This table has three columns. On a narrow screen, use touch, a mouse or the keyboard to move horizontally inside the table region without making the page itself overflow.

Jewellery product fields, AI-assisted work and human confirmation responsibilities
Product fieldWork AI may assist withWhat a person or system of record must confirm
Material and finenessTurn approved facts into titles, bullet points and bilingual copyThe actual 18K/750, platinum, plating or other specification
Stone originApply approved “natural” or “laboratory-grown” wording consistentlyThe origin of the item; never infer this from an image
CertificatePlace an approved laboratory, report number and link on the pageAuthenticity, the match between report and item, and what may be made public
Price and availabilityGenerate page or notification drafts from connected dataCurrent price, currency, stock, preorder/backorder state and quotation validity
ImagesProduce a clearly labelled contextual visual or layout mock-upWhether the main product photograph is accurate; a composite must not pose as evidence of the physical item

The product page, structured data and product source must describe the same item

Google explains that a product page can use Product structured data, submit product data through Merchant Center, or use both. Providing both can maximise eligibility for different shopping experiences and help Google understand and verify the data. Supported Merchant Center fields include product ID, title, description, image, brand, identifiers, material, price and availability. These machine-readable values should come from the same approved product record, not from a parallel version rewritten solely for search.

For jewellery sites, a recurring risk is that the page says “in stock” while the source says “out of stock”; the gold-linked price or promotion changes while structured data and checkout do not; or a listing sends a shopper to a different default variant. Google’s Merchant Center documentation requires price, currency and availability to agree across the landing page, structured data, product data and checkout. Mismatches can lead to product disapproval.

For fixed-price products, use one inventory and pricing source to update visible copy, Merchant Center and structured data. If an item is enquiry-only or individually quoted, do not invent a fixed price or real-time stock state. State that a quotation is required, what affects it, how long it remains valid, the selectable specifications and the contact route. Standard Merchant Center free listings require price and availability, so an enquiry catalogue should not masquerade as directly purchasable product data.

AI-generated product copy and images still need labels and human approval

Google Merchant Center currently requires product images created with generative AI to contain IPTC DigitalSourceType metadata, such as TrainedAlgorithmicMedia, and to preserve relevant embedded tags. If the tool has not added the required metadata, add it before submission. Titles and descriptions created with generative AI must use structured_title and structured_description respectively, each with digital_source_type: trained_algorithmic_media and the required content text.

Those are Merchant Center submission requirements; they do not mean that AI-generated content receives better ranking. CLEARCRAFT’s separate operational recommendation is to add human approval before publication. A reviewer should verify material, natural versus laboratory-grown classification, certificate details, dimensions, price, availability and the nature of each image. A contextual composite should also be described to customers as an illustration and accompanied by an accurate product image. The official data requirement and this customer-communication recommendation should remain distinct in the record.

GEO should reduce contradictory data, not promise rankings

Google’s official guidance says that AI Overviews and AI Mode remain rooted in the usual Search indexing, quality and ranking systems. Merchant Center and product information can help products become understandable in Google Search and appropriate AI responses, but meeting technical and content requirements still does not guarantee crawling, indexing, serving or citation. Google also says there is no special structured-data type required for generative AI search.

The practical GEO value for a jeweller is therefore not creating a separate page for every long-tail wording. It is keeping one product identity and one set of verified facts consistent across the product page, data source, structured data, enquiry form and staff response. That reduces ambiguity for search systems, sales staff and customers at the same time.

An enquiry workflow that does not let AI quote on its own

For high-value, customised or price-sensitive jewellery, a website can use AI or a form to collect a customer’s needs while reserving commitments for authorised staff:

  1. Start the enquiry from a specific product page and preserve the product ID, variant, language and page URL.
  2. Ask for intended use, budget range, ring size or dimensions, natural or laboratory-grown preference, timing and contact route. Do not infer information the customer did not provide.
  3. Answer routine specifications only from approved data. Route certificate status, authenticity, live stock, gold-linked price, discount and delivery date to “human confirmation required”.
  4. Have an authorised member of staff check the physical item, certificate, stock and current price, then issue a versioned quotation with an expiry time and accountable owner.
  5. Keep revisions under the same enquiry ID and record which fields came from the system and which were confirmed by a person, so that channels do not contradict one another.

This is not a Google ranking rule. It is a CLEARCRAFT risk-control recommendation designed to hand a search-originated enquiry to a person without mixing public product facts, unverified inferences and commercial commitments.

Minimum pre-publication check

  • Does every product and variant have a stable ID that resolves to the correct product page and selectable variant?
  • Does visible copy clearly state the material, natural or laboratory-grown status, principal specifications and supply method?
  • Has a person verified certificate details, and have private fields been kept out of generative-AI tools?
  • Do price, currency, availability, structured data, Merchant Center and checkout agree?
  • Are AI-generated images, titles and descriptions labelled as Merchant Center requires?
  • Does the enquiry or quotation flow include a clear human approval point, validity period and accountable owner?

Completing these checks does not create a Google or AI-platform guarantee. It creates product information that is easier for customers to understand, systems to verify and the business to maintain.

PRIMARY OFFICIAL SOURCES · CHECKED 5 NOV 2026

Primary official sources

  1. Google Search Central: Optimizing your website for generative AI features on Google Search
  2. Google Search Central: Introduction to Product structured data
  3. Google Search Central: Product variant structured data
  4. Google Merchant Center Help: Product data specification
  5. Google Merchant Center Help: Supported structured data attributes and values
  6. Google Merchant Center Help: AI-generated content
  7. Google Merchant Center Help: Tips to optimise your product data
  8. Google Merchant Center Help: About landing page requirements

Sources checked: 5 November 2026. Platform requirements in this article are based only on the public Google documentation above. CLEARCRAFT workflow recommendations are identified as recommendations in the body. The article uses no client data, private jeweller inventory, private certificate, unauthorised case study or outcome figure.

Corrections and limitations: this article is not an endorsement by Google, Gemini or another platform, and it does not guarantee ranking, citation, traffic or revenue. Verifiable factual corrections follow the editorial and corrections policy.