Jeweler AI Search Citations on Squarespace

The custom-engagement-ring shape is the higher-AOV opening¹. A buyer searching "custom engagement ring Charleston lab-grown" is a buyer with a budget in the $2,000-$8,000 range, an active stone preference, and a deliberate choice not to walk into the local Kay or Jared. Three to five independent jewelers in Charleston are credible candidates for the citation, the buyer's question stacks five constraints, and the marketplace head terms ("engagement rings", "lab grown rings") offer no path to that buyer. The independent jeweler who writes a 1,200-word page titled "Custom engagement rings, Charleston SC — lab-grown stones, hand-forged settings" with the answer-first 134-167 word lead is the only candidate the engine has to cite on that exact constraint stack.

The handmade-jewelry shape is the higher-volume opening at a lower per-piece value. A buyer searching "handmade silver hammered cuff" is a buyer somewhere in the $80-$400 band browsing across Etsy, Instagram, and a handful of independent maker sites. Etsy owns the marketplace head term on Google, but the page Etsy returns is a filtered list of 400 cuffs — the AI engine cannot extract a fluent paragraph from that list. The jeweler with a Squarespace product page that opens "Hammered Sterling Silver Cuff — 1.5 inches wide, 28 grams, hand-forged from recycled sterling, made-to-order in 3-4 weeks by Linden Bay in Charleston" is the page the engine can quote. The engine quotes it because the page supplied the paragraph.

The first filter is crawler access. Squarespace's AI exclusion checkbox⁷ ships unchecked by default, but a meaningful share of jeweler sites toggled it on after 2024-era posts that conflated training crawlers (which scrape content into training sets) with retrieval crawlers (which fetch pages in real time so engines can cite). The exclusion box treats both as one, and toggling it on cuts off the citation path entirely. The first audit step is verifying the box is unchecked and that ChatGPT-User, OAI-SearchBot², and PerplexityBot³ can all reach every top product page the studio wants cited.

The second filter is passage shape. AI engines prefer product descriptions where the first 134-167 words directly answer the customer's full constraint stack without requiring a click. A jeweler product page targeting "handmade silver hammered cuff" should open the description with a bolded passage stating the material (sterling silver, recycled), the style (hammered, hand-forged), the item type and dimensions (cuff at 1.5 inches wide, 28 grams), the maker's hand (made-to-order in 3-4 weeks), and the city or studio name. Below the lead the page expands with care instructions, resizing policy, and the maker's process. The lead is what the engine quotes; the expansion is what the engine reads to confirm the product is real.

The contrast with an Etsy listing or a Blue Nile category page is the strategic point. Etsy listings are field-driven and front-end-filtered — title, tags, attributes, a short paragraph. AI engines can read all of it, but they cannot extract a fluent passage from a tag set. A Squarespace product page that puts the same information into a 134-167 word answer-first lead reads as a citable passage and the engine quotes it; the same information distributed across Etsy's listing fields reads as a database row and the engine cites the marketplace instead.

Four details inside the example matter for AI citation⁵. Specific material grade ("recycled sterling silver" beats "silver"). Specific dimensions in the lead ("1.5 inches wide, 28 grams" — both the dimension and the weight). Specific maker timing ("made to order in 3-4 weeks" — buyer plans around it). Named city ("Charleston, South Carolina" — closes the location constraint on local queries). The engines reward this specificity because the customer's query asked for it, and the page that supplies it directly is the page the engine has the highest confidence quoting.

The discipline that makes the knowsAbout array work is honesty. List the techniques the studio has actually shipped pieces in, not the techniques the studio aspires to. A maker who works primarily in hand-forged silver and lost-wax cast gold lists "hand forging", "lost-wax casting", "recycled silver work", and "recycled gold work" — not the full craft glossary. The engine reads the array against the product pages, finds matching pieces tagged with the same material and technique vocabulary, and the citation graph closes around a verifiable maker with a verifiable specialism rather than an anonymous brand.

 <script type="application/ld+json"> { "@context": "https://schema.org", "@type": "Person", "@id": "https://yourstudio.com/#maker", "name": "Linden Bay", "jobTitle": "Principal Goldsmith", "url": "https://yourstudio.com/about/", "worksFor": { "@type": "JewelryStore", "name": "Linden Bay Jewelry" }, "hasCredential": "GIA Graduate Gemologist", "knowsAbout": [ "Lost-wax casting", "Hand engraving", "Lab-grown diamond settings", "Recycled gold and silver work", "Art Deco restoration", "Custom engagement rings" ], "sameAs": [ "https://www.instagram.com/lindenbayjewelry", "https://www.gia.edu/alumni/linden-bay" ] } </script> 

The sameAs links matter. AI engines use them to disambiguate one named maker from another, and a jeweler with an Instagram account plus a GIA Alumni listing is significantly easier to confidently attribute than an anonymous name on a single domain. The GIA Alumni link is the highest-leverage one because it functions as a verification path confirming the maker is a real credentialed gemologist, which is the trust signal AI engines reward most heavily for buyer-decision queries on stones. For a non-gemological studio (a goldsmith working in textured bands and cuffs, for instance), the equivalent verification path is a professional association or a published-feature link — anything that closes the entity loop with a third-party-verifiable claim.

The mechanism. Google's image search and Perplexity's image-citation cards both extract caption and creditText from ImageObject⁶ when deciding which image to surface on a material + style image query. A jewelry product page with the hero shot plus four detail angles (profile, hand-detail, scale-on-finger, material texture) and an ImageObject node on each surfaces on image-flavoured queries that the bare-image product page cannot. The caption field carries the descriptive sentence ("Hammered sterling silver cuff, 1.5 inches wide, profile view"), the creditText carries the maker name ("Linden Bay, Charleston SC"), and the engine has the metadata it needs to render a confident image-citation card.

The Squarespace gap. Squarespace Commerce auto-emits the basic image URL in the Product JSON-LD block, but does not auto-emit ImageObject nodes with caption, creditText, copyrightHolder, and contentLocation. A supplemental schema block in Page Settings > Advanced > Page Header Code Injection on each top product page adds one ImageObject node per featured image — typically three to five per product — and closes the image-citation gap. The full ImageObject field list lives on the schema pillar; the jeweler-specific application is the install layer above. Combined with descriptive alt text in the Squarespace image editor (the alt-text field on each product image), the ImageObject markup is what graduates a jewelry product page from "visible to Google Images" to "citable by ChatGPT and Perplexity on material + style image queries".

The query list is the leverage¹. Generic queries ("jewelry store [city]", "engagement ring") are easy to track but Etsy and the chains already own them; material + style + city queries ("handmade silver hammered cuff", "custom engagement ring Charleston lab-grown", "Art Deco emerald cocktail ring") are harder to track because they require specific phrasing — but those are the queries where citation movement actually shows up. The right list mixes one or two broader queries with eight to twelve material-stacked queries the studio genuinely wants to be the answer for. Update the list quarterly as the studio's catalog mix evolves.