yoreh® AI Visibility Audit

Good news first. yoreh.co is fully readable by every major AI crawler we tested (GPTBot, PerplexityBot, ClaudeBot, plus a standard browser baseline). The site serves the same complete HTML to all of them. Shopify is also auto-publishing the new AI agent discovery files (llms.txt, agents.md, an agentic sitemap, an MCP endpoint), which puts the store on the right side of a quiet platform shift that happened in the last six months. The technical foundation is one of the strongest you can have for AI visibility right now.

The work is on the content layer. The homepage has no H1, the visible H2 tags are placeholder text ("landing page", "landing page2", "Landing page 3"), and there's no tight brand definition near the top of the HTML for AI to extract on "what is yoreh" type questions. The public catalog also exposes duplicate or test products that quietly hurt brand authority. None of this requires a rebuild. All of it is editable in Shopify admin in a few hours.

The three things that matter

• The hardest part is already solved. Most stores fail the basic crawlability test (the homepage is JavaScript only and serves a blank shell to bots). yoreh passes it cleanly. AI sees the catalog, the descriptions, the reviews, and the editorial.
• You're early on AI agent commerce. Shopify quietly shipped llms.txt, agents.md, and an MCP endpoint at /api/ucp/mcp. yoreh is exposing all of them. When AI shopping agents (OpenAI, Perplexity, ChatGPT actions) start routing buyers, stores with these files are on the discoverable side.
• The remaining work is a content sweep plus a slow authority play. Most fixes are editable in Shopify admin in a few hours. The authority piece (third-party reviews, founder content, press) runs in parallel over a quarter. Detailed in Section 11.

The goal was to answer one practical question: when an AI assistant tries to read yoreh.co, what does it actually see, and is that the version of the brand you'd want it to cite? We bypassed JavaScript entirely (which is what AI crawlers do) and read the raw HTML each crawler is served.

Every AI crawler we tested received a complete server rendered HTML payload, identical in size and content to what a human browser receives.

This is the single most common point of failure across the AI visibility audits we run, and roughly half the stores we look at fail here. The most common cause is a JavaScript only render where the bot gets back an empty page shell. yoreh ships server rendered HTML, so the moment a model crawls the site the brand's catalog, descriptions, reviews, and editorial are visible to it.

In the last six months Shopify began publishing a set of files designed for large language models and AI shopping agents. yoreh is exposing the full set:

What this means in practical terms: when an AI shopping agent looks up "alternative silver jewelry brands" or "buy a 925 silver ring under 200 dollars", stores with these files are reachable. Stores without them aren't yet, and the spec is moving fast enough that retrofitting later is more work than starting in this position.

This is the single biggest fix in the audit. AI crawlers anchor extraction to heading tags. They use the H1 to understand the page topic and the H2s to understand the section structure. On the yoreh homepage today the crawler sees this:

The homepage <title> and meta description are well written. The visible text underneath is mostly product cards and currency selectors. There's no 2 sentence brand definition near the top of the HTML, which is the format AI most reliably lifts when answering "what is yoreh" or "is yoreh legit" or "yoreh jewelry" queries.

What to add (suggested)

<h1>Alternative 925 Silver Jewelry, Handcrafted</h1>

<p>yoreh® is an independent jewelry brand making handcrafted
925 silver and gold vermeil pieces in limited runs, including
the only ring set with a genuine fragment of lunar meteorite.
Designed in Hong Kong. Worn by people who don't blend in.</p>

Section H2s should describe the section: New Arrivals, The Lunar Meteorite Series, What People Say, Shop All Silver. This is a 30 minute job in Shopify admin, and it's the highest leverage single change in this audit.

The product sitemap currently exposes pages that look like duplicates or work in progress:

• /products/boyy-bracelet-silver-copy
• /products/carre-bracelet-silver-copy
• /products/copy-of-tycho-ring-silver
• /products/hitomi-pendant-silver-copy

And the pages sitemap exposes:

• /pages/blog-post
• /pages/blog-post-1

None of these are linked from navigation, but they're in the sitemap, which is what AI crawlers prioritise. When a model crawls the catalog and sees four -copy products and two blog-post-N placeholders, the implicit signal is that the brand ships duplicates and runs unpolished pages. Authority is one of the three things AI uses to decide whether to cite a source. Small leaks like this compound.

Schema markup (JSON-LD) tells AI what kind of content a page is. yoreh has the basics. The high value extras are missing.

What's there

What's missing

Jewelry buyers ask AI assistants exactly these questions before they buy. Right now those answers live elsewhere, on Reddit threads and competitor blogs. Adding FAQPage and HowTo schema with 8 to 12 well written Q&As is the cheapest way to redirect that citation traffic to yoreh.

The blog index lists four sections (the-archive, art-of-intent, a-chat-with-fin, moon-rocks). The lunar meteorite story is genuinely original, citable content (the kind of unique angle AI loves to cite because it can't get it elsewhere). The category needs more.

For an alternative jewelry brand, AI assistants prefer editorial that contains:

• A named author with credentials. Fin as designer and maker, with a short bio block. This is an authority signal AI weights heavily.
• Specific data with attribution. Percentage of recycled silver in each cast, the sourcing partner for the meteorite fragments, year of authentication, weight in grams. Naked statistics get deprioritised. Numbers with named sources get cited.
• Structured how-to sections. Sizing, care, gift selection, storage, polishing. Each as a numbered list with H2 anchors so AI can extract a single section without reading the whole post.
• Self contained product stories. The meteorite series story should exist as one page that answers "what is the lunar meteorite ring" completely, end to end, with the H1 phrased as the question.

Three to five well structured pieces here would meaningfully change which sources AI pulls from when buyers ask category questions.

AI systems don't just cite the most relevant page, they cite the most credible one. yoreh's content layer is fixable in weeks. The authority layer takes longer because it requires third-party signal that doesn't live on yoreh.co at all.

The yoreh.com leak

The .com domain (yoreh.com, not the live .co) returns an SSL certificate error for anyone who types it. Any buyer who hears the brand by ear, defaults to .com, and lands on a browser security warning. This is a five-minute fix at the registrar or DNS layer: either install a cert and 301 redirect to yoreh.co, or release the domain. Either way, close the leak.

Brand confusion risk: Yordy.co

Yordy.co is a separate gothic fashion jewelry brand. Different products, different price tier, different aesthetic. The names differ by one letter, both are jewelry, both lean alt. Yordy has 5,645 Judge.me reviews and a much larger review surface. When a buyer asks an AI "is yoreh good quality", models can confuse the entities and pull from Yordy's reviews instead of yoreh's actual signal.

The fix is not engagement with Yordy. It's making yoreh's first-person identity unmissable: a clear definition block, a named founder with credentials, a stated origin (Bali workshop), and structured product stories. The clearer the entity, the harder it is for AI to conflate.

You said you weren't sure what queries to target. Here's how to think about it. AI search visibility splits into four query types, ordered by how fast yoreh can realistically win each.

Order of operations: branded first (three weeks of on-site work, fully controllable), educational and local in parallel (one to two quarters), listicle as the long-haul play. The lunar meteorite series is the easiest unique angle to lead with. There are very few credible sources globally for "jewelry with real lunar meteorite", so AI is hungry for citable material there and yoreh has it natively.

Ordered by impact divided by effort. The first four are under an hour combined and close most of the on-site gap. The last two are the slower authority play that runs in parallel.

• Replace homepage H1, H2s, and add a brand definition block. The single highest leverage change. Real H1, descriptive section H2s, two sentence brand definition above the first product carousel. 30 min, Shopify admin
• Archive duplicate and test products and placeholder pages. Closes the authority leak in the catalog. 5 min, Shopify admin
• Fix the yoreh.com leak. Either install a cert and 301 redirect to yoreh.co, or release the domain. Stops the browser security warning anyone typing the .com hits today. 5 min, registrar / DNS
• Add explicit AI bot Allow rules to robots.txt. Defensive hygiene. Add explicit Allow blocks for GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Google-Extended, Applebot-Extended. Implicit allow becomes explicit, so future Shopify default changes can't quietly break crawl access. 10 min, Shopify admin (robots.txt.liquid)
• Add a structured FAQ page with FAQPage schema. 8 to 12 Q&As covering 925 silver, gold vermeil, lunar meteorite authenticity, sizing, care, returns, ethical sourcing, made to order vs in stock. Each Q&A wrapped in JSON-LD. 2 hours
• Convert ring size guide, care guide, and product guide to HowTo schema. Number the steps, mark them up with HowTo JSON-LD. 1 to 2 hours total
• Add Person schema to Meet the Team page. JSON-LD Person blocks for Fin and the team, with sameAs links to LinkedIn and Instagram. AI uses author / founder identity signals to verify brand authority. 30 min, theme edit
• Run a citation baseline. Test 10 target queries on Perplexity, ChatGPT (web), and Google AI Overviews. Log who's cited today. This is the "before" measurement we'll re-run in 4 weeks to see what moved. 1 hour
• Set up third-party review collection. Trustpilot or Judge.me. On-site reviews already exist (31 on the Libelula pendant alone), but there's no third-party review surface for AI to cite when buyers ask "is yoreh good quality". Email last 6 months of customers to seed it. 1 hour setup, then ongoing
• Editorial expansion. Three to five deeper pieces with named author, attributed data, and structured sections, focused on the unique stories (meteorite series, sourcing, design process). 4 to 6 weeks of cadence
• Founder content and press play. Three podcast or interview placements per quarter (jewelry founder podcasts, alt-fashion verticals, sustainable craft outlets). Plus targeted PR for jewelry roundups (Hypebeast, Highsnobiety, Cool Hunting). The quarter-long authority lever. Ongoing, 1 quarter to first results

We've been running this same audit across our own clients and a few friends' brands. yoreh's starting position is unusually strong. Most stores at this stage fail the basic crawlability test. yoreh passes it cleanly, and the Shopify AI agent layer is already configured.

What's left is a content sweep plus a slower authority play, not a rebuild. The four highest-leverage on-site fixes are under an hour of admin work combined and close most of the citation gap. The schema and editorial layer earns the rest. Authority (third-party reviews, founder content, press) runs in parallel over a quarter, because lead times there are long.

This audit is a snapshot of one moment in time. AI search changes fast: schema models shift, new crawlers appear, citations rotate every few weeks. A static document can tell you what to fix. It can't tell you whether the fixes are working. That's where the measurement layer comes in.

What yoreh can ship this week (no tooling needed)

• Fix the four highest-leverage on-site items (Findings 3 and 4, plus the yoreh.com SSL leak and explicit robots.txt allows). Under an hour of admin combined.
• Start the founder content and review-collection work in parallel. Lead times are too long to leave them for later.

The measurement layer: Node AI's AI Visibility platform

For the prompt and scan side, we use our own AI Visibility platform. It's the same tool we run for our other clients, and it's how we know whether each fix is moving the needle in AI search.

The platform replaces the "test 4 queries on 4 platforms every Monday" workflow that most brands try to run by hand. It's the same operating layer that lets us tell our other clients exactly which AI model invented which fact about them, and which fixes shifted recognition by which percent.