AI SEO / AEO Audit — Dear Eden (shopdeareden.com)

Prepared 2026-08-26. Context from business-analysis.md (26-section brand study) plus live technical checks of robots.txt, /agents.md, /llms.txt, /.well-known/ucp, the sitemap, raw HTML schema, and 9 representative web-search visibility probes across ChatGPT/Perplexity-style query types. All findings verified directly against the live site unless flagged.

---

0. Executive Summary

Dear Eden is unusually well-positioned for agentic / answer-engine commerce — it already implements the Universal Commerce Protocol (/agents.md, /.well-known/ucp), allows all AI crawlers in robots.txt, and carries clean Product/Organization/WebSite schema. That is ahead of 95% of micro-brands.

But it is not optimized for being cited as an answer by ChatGPT, Perplexity, Gemini, or Google AI Overviews. Why:

1. No extractable answer content exists. The site is a pure storefront — no blog posts, no FAQ section, no definition/comparison blocks. AI engines cite content, and Dear Eden has published none. Visibility probes confirm the store appears only for branded/tail-term queries (it ranks #1 for itself) and is absent from every non-branded, category-level, "best brands," and "how to choose" query we tested.
2. /pricing.md and /pricing.txt both 404 — the single most important file for an AI buying agent is missing. Agents evaluating products on a buyer's behalf cannot read structured pricing.
3. /llms.txt is a copy of /agents.md (4 lines differ). It is agent instructions, not the llmstxt index that points AI at your best pages. This is a misfire.
4. No AI-citable authority signals — no third-party press, no Wikipedia, no review-site presence, no public statistics. The brand is a Junior Style Awards 2026 nominee (a genuinely citable third-party signal) but that fact lives only on an external vote page and is not surfaced as structured content on-site.

Bottom line: Fix the cheap structural gaps (pricing file, llms.txt, FAQ schema) now — they are near-zero effort and directly help the agent-commerce channel the brand already invested in. But real AI citation visibility requires publishing answer content (FAQ + buying-guide articles), which is a net-new content effort the brand currently has zero of.

---

1. AI Visibility Snapshot (Probed Queries)

Tested 9 queries across the query-type matrix. Method: web_search probes (stands in for ChatGPT/Perplexity/Gemini retrieval). "Cited" = the store or brand appears in results.

#	Query	Type	Dear Eden Cited?	What Shows Instead
1	matching mother daughter dresses	category tail	No	Pinterest, TikTok, Temu, Etsy, Amazon
2	Dear Eden shopdeareden matching outfits	branded	Yes (#1)	Homepage + Instagram
3	best mother daughter matching outfits designer	best-of	No	Pinterest, PatPat blog, romanceuniversity.org
4	French linen mother daughter dresses heirloom	long-tail	No	Pinterest, AliExpress, Autumn Moon, dresslikemommy
5	Dear Eden brand review mommy and me	branded/rep	Yes (#1)	Homepage, Instagram (33K), Facebook
6	what are the best mommy and me clothing brands 2026	best-of/listicle	No	The Bump, Pinterest, TikTok (no Dear Eden)
7	mommy and me matching mother daughter	generic	No	Pinterest/TikTok/Amazon/Temu/Etsy
8	French linen mother daughter dresses heirloom	niche	No	Same as #4
9	Dear Eden matching swim collection	branded	Yes	Instagram reel, homepage

Read: Dear Eden is only retrievable on branded or exact-tail queries ("Dear Eden…", "shopdeareden"). On the 6 non-branded queries (the ones that actually drive discovery), it is never cited. Competitors cited where Dear Eden is not: PatPat (blog + commerce), The Bump (editorial listicle), Rylee + Cru / Jamie Kay class (premium niche), and a long tail of Pinterest/Etsy/Temu/AliExpress marketplaces.

This matches the business analysis: ~700 US-only visits/month, no marketplace presence, no content engine. AI engines behave exactly like humans here — with nothing to index beyond product pages, there is nothing to cite.

---

2. AI Crawler Access (robots.txt) — PASS

Verified live. The store does not block any AI crawler. User-agent: → Allow: /, with only transactional/private paths disallowed (/admin, /cart/, /checkout, /orders, /account, AJAX surfaces, Shopify-internal /services, /sf_). Notably the robots.txt itself invites agents:

# Agent instructions: https://shopdeareden.com/agents.md
# UCP discovery: https://shopdeareden.com/.well-known/ucp
# Agents should use UCP/MCP for catalog, cart, and checkout.

GPTBot, PerplexityBot, ClaudeBot, Google-Extended, Bingbot are all permitted. No action needed. (Contrast with the common mistake of blocking training crawlers — not done here. Sitemap is declared: Sitemap: https://shopdeareden.com/sitemap.xml.)

---

3. Machine-Readable Agent Files

File	Status	Notes
/agents.md	200 (live)	Correct UCP + Shop-skill agent instructions. Well-formed.
/.well-known/ucp	200 (live)	Valid UCP merchant profile (versions 2026-04-08, 2026-01-23; shopping MCP endpoint).
/llms.txt	200 — but WRONG CONTENT	It is a 4-line-different copy of /agents.md. This is the agent-instructions file, not the llmstxt index.
/pricing.md	404	Missing. Should list structured pricing for AI buyers.
/pricing.txt	404	Missing (same gap).
/.well-known/ucp MCP endpoint	live	POST /api/ucp/mcp available for agent-driven commerce.

3a. Fix /llms.txt (HIGH, near-zero effort)

llms.txt (per llmstxt.org) should be a short index that tells AI systems what the site is and which pages matter — not a rehash of agent plumbing. Replace it with a proper index (a deployable corrected version is provided at aeo-deploy/llms.txt). It should:

• Name the brand and one-sentence definition.
• Link the key pages (home, collections, about, shipping, contact, and crucially the new /pricing.md).

3b. Add /pricing.md (HIGH, near-zero effort)

This is the single highest-leverage AEO fix. An AI buying agent evaluating "mother-daughter linen dress sets under $300" cannot read JavaScript-rendered prices reliably; a markdown file is trivially parseable. A deployable version with the full current catalog pricing is provided at aeo-deploy/pricing.md. Keep it updated whenever prices change (stale pricing is worse than none).

---

4. Structured Data (Schema) — Mixed

Verified by fetching raw HTML (curl) and grepping application/ld+json — not cleaned/markdown text, which would have stripped the script tags.

Page	Schema present	Types found	Gap
Homepage	Yes (2 blocks)	Organization, WebSite (+SearchAction)	No FAQPage, no hero-product Product block
Product (Louise midi)	Yes (2 blocks)	Product, ProductGroup, Brand, Offer	No aggregateRating in JSON-LD despite 23 reviews on-site
About	None found	—	No AboutPage/Organization/Person(founder) schema
Blog (/blogs/news)	None (and empty)	—	No posts published at all
Shipping/Contact	None found	—	No structured policy schema

Details:

• Product schema is solid: price ("price":"175.00", priceCurrency:USD), Offer, Brand, ProductGroup all present and correct. Good foundation.
• Missing AggregateRating: Judge.me shows 23 reviews / 5.00 average on the homepage and per product, but no aggregateRating or Review JSON-LD is emitted. AI engines therefore see the product as "unreviewed." Add AggregateRating (+ optionally Review) to product JSON-LD — or enable a Judge.me/theme option that emits it.
• About page has no schema — a missed chance to establish the founder (Samantha Links) and Louisiana HQ as a citable entity.
• No BreadcrumbList schema on collection/product pages (AI uses breadcrumbs to understand category hierarchy).
• No FAQPage schema anywhere — and there is no FAQ content to back it (see §5).

---

5. Content Extractability — the Core Gap (CRITICAL)

AI engines cite passages, not pages. Dear Eden currently publishes zero extractable answer content:

• No blog / editorial content. /blogs/news returns 200 but contains no posts. There is no "how to choose a mommy-and-me size," "what is French linen," "mother-daughter outfit ideas for photos," or "gifting guide" content — the exact long-tail, high-intent queries AI fans out to.
• No FAQ section on any page (home, product, policy). The only "FAQ" reference on-site is inside the UCP agent tool schema (an agent-facing tool description), not user-facing FAQ content.
• No definition / comparison / pros-cons blocks. Product copy is brand-voiced ("designer worthy without the luxury markup") but not structured as self-contained 40–60 word answers to "What is…?", "X vs Y?", "How to…?" queries.
• Hero copy lacks a plain-language definition. The homepage leads with "THE LOUISE DRESS COLLECTION / IN YOUR SIZE & HERS" — visually strong, but there is no first-paragraph sentence a model could extract as "Dear Eden is a ___ brand that ___."

Why this blocks AEO: Google's AI Overviews correlate with classic ranking, and ChatGPT/Perplexity/Gemini all pull from content that answers the query. With only product pages indexed, Dear Eden can only ever appear for product-name or brand queries — exactly what the §1 probes showed.

Recommended content to publish (priority order)

1. FAQ page (and FAQPage schema) covering: shipping & free-shipping thresholds, sizing across generations, fabric care (French linen), "what is mommy-and-me," gifting, returns. ~10–15 natural-language Q&As. This is the fastest path to FAQ-schema eligibility and featured-snippet extraction.
2. Founder / brand story article (the About content, expanded into extractable prose with the Junior Style Awards 2026 nomination stated plainly). Enables "who is Dear Eden" citations.
3. 2–3 buying-guide articles: "How to Choose Matching Mother-Daughter Outfits," "French Linen vs. Cotton for Kids' Special-Occasion Dresses," "Mother-Daughter Photo Outfit Ideas." These target the non-branded queries that currently cite competitors (PatPat blog, The Bump).
4. Comparison content: "Dear Eden vs. [Rylee + Cru / Jamie Kay]" style honest comparison (comparison tables get cited ~33% of the time per GEO research).

---

6. Authority & Presence Signals — Weak

Per the Princeton GEO study, the biggest citation boosts come from citations/sources (+40%), statistics (+37%), and expert quotes (+30%). Dear Eden has almost none of these as citable on-site or third-party assets:

• Third-party presence: Only Instagram (33K followers, confirmed) + Facebook (415 likes). No TikTok, Pinterest, YouTube, Reddit, Quora, Wikipedia, or review-site presence. Pinterest and TikTok are especially damaging gaps for a visual, maternal, gifting brand — and they are exactly where AI Overviews pull "how-to" and inspiration content from.
• Press / awards: Junior Style Awards 2026 Nominee is a real, citable third-party signal — but it lives only on juniorstyle.net, not surfaced as structured content or linked prominently on-site.
• Statistics: None published (e.g., "23 five-star reviews," "small-batch of N units," "ships to 200+ countries"). Adding dated, sourced numbers to product/about pages lifts citation likelihood.
• Author attribution: Founder Samantha Links is named on About, but there is no author bio/credentials block that AI can attribute quotes to.
• Reviews not machine-readable: Judge.me reviews render client-side; the 23-review average is not in product JSON-LD (see §4).

Presence actions (non-content, mostly free):

• Claim/optimize Pinterest and TikTok (highest-leverage for this demographic) and link them in the footer (currently only IG + FB).
• Pursue a Wikipedia mention or at minimum ensure the Junior Style Awards nod is linked from the homepage/about.
• Encourage authentic Reddit/Pinterest participation in mommy-and-me and wedding/photo communities (no astroturfing — real participation only).

---

7. Agentic Experience Readiness — STRONG (differentiator)

This is where Dear Eden is genuinely ahead:

• UCP + /agents.md + Shop skill implemented — an AI agent can discover the catalog, price, cart, and (with human approval) checkout via standardized endpoints. Most micro-brands have none of this.
• Semantic HTML present: <main id="MainContent" role="main">, <nav>, proper <h2> hierarchy (10 on homepage), image alt attributes, aria labels on interactive elements. Agents parsing the DOM or accessibility tree will get clean structure.
• Visible, indexable pricing & policies on public pages (not behind login walls) — good for both humans and agents.
• Clean canonical + OpenGraph tags on the homepage (good for shareability and entity clarity).

Caveats: Some content (reviews, possibly variant pricing) is client-rendered. For agent reliability, the /pricing.md file (§3b) and aggregateRating JSON-LD (§4) close the gaps so agents don't have to execute JavaScript to learn price or trust signals.

---

8. Prioritized Recommendations

Tier 1 — Do now (hours, near-zero cost, directly helps the agent channel already built)

1. Publish /pricing.md (deployable file provided: aeo-deploy/pricing.md). Add /pricing.txt mirror. Link from sitemap + footer.
2. Rewrite /llms.txt as a proper index (deployable file provided: aeo-deploy/llms.txt). Stop duplicating agents.md.
3. Add AggregateRating (+Review) to product JSON-LD so the 23 five-star reviews become machine-readable trust signals.
4. Fix the broken footer About link flagged in business-analysis.md (/pages/about-the-brand 404 → /pages/about-us). Broken policy/about links erode AI trust signals.

Tier 2 — Publish answer content (weeks, the real AEO lever)

1. FAQ page + FAQPage schema (shipping, sizing, fabric, gifting, returns). Fastest route to snippet/FAQ extraction.
2. Founder/brand-story article with the Junior Style Awards 2026 nomination stated plainly + author bio.
3. 2–3 buying-guide / comparison articles targeting the non-branded queries competitors currently win (The Bump, PatPat blog, Pinterest).
4. Add BreadcrumbList + AboutPage/Person schema to collection, product, and about pages.

Tier 3 — Authority & presence (ongoing)

1. Launch Pinterest + TikTok and link in footer (currently IG + FB only).
2. Surface the Junior Style Awards nomination on-site (badge + linked mention) and pursue Wikipedia/editorial mentions.
3. Publish dated statistics ("23 five-star reviews," "ships to 200+ countries," "small-batch artisan production") across product/about pages.
4. Begin DIY AI-visibility monitoring: monthly manual check of the §1 query set across ChatGPT, Perplexity, Google AI Overviews; log citations in a sheet. (No AI-specific Search Console report exists — third-party tools like Otterly/Peec/ZipTie are optional.)

---

9. What NOT to do (per Google's AI optimization guide)

• Do not write separate "for AI" content variants — that risks scaled-content-abuse flags. The FAQ/guides above should serve humans first.
• Do not block GPTBot/PerplexityBot/ClaudeBot/Google-Extended (currently allowed — keep it that way).
• Do not keyword-stuff (active −10% AI-visibility penalty in GEO study).
• Do not leave pricing/contact behind JS-only rendering for agents — the /pricing.md file prevents this.

---

10. Citations & Sources

• https://shopdeareden.com/ — homepage: hero, navigation, collections, Judge.me badges (23 reviews / 5.00), Junior Style nominee, footer social (IG + FB), Shopify/Cloudflare, semantic <main>/<h2> structure, canonical + OpenGraph tags.
• https://shopdeareden.com/robots.txt — verified: User-agent: * Allow /; no AI crawler disallowed; UCP/agents.md referenced; sitemap declared.
• https://shopdeareden.com/agents.md — verified live (UCP + Shop skill agent instructions).
• https://shopdeareden.com/llms.txt — verified live but content duplicates agents.md (4 lines differ); not a proper llmstxt index.
• https://shopdeareden.com/.well-known/ucp — verified live (UCP merchant profile, v2026-04-08 / 2026-01-23, shopping MCP endpoint).
• https://shopdeareden.com/pricing.md — verified 404 (gap).
• https://shopdeareden.com/pricing.txt — verified 404 (gap).
• https://shopdeareden.com/products/womens-louise-french-linen-lace-midi-dress-eden-stripe-natural — verified Product/ProductGroup/Brand/Offer JSON-LD, price 175.00 USD, no aggregateRating.
• https://shopdeareden.com/pages/about-us — verified: founder Samantha Links, Louisiana HQ, artisan partners India & China; no schema block found.
• https://shopdeareden.com/blogs/news — verified HTTP 200 but contains no published posts.
• https://juniorstyle.net/vote-in-the-junior-style-awards-2026/ — Dear Eden listed as Junior Style Awards 2026 Nominee.
• https://instagram.com/shopdeareden — 33K followers (confirmed via search result snippet).
• https://facebook.com/shopdeareden — 415 likes (confirmed via search result snippet).
• Business context: /root/cro-reports/shopdeareden/20260826-102324/business-analysis.md (26-section brand study; ~700 US-only visits/mo; 23 SKUs; $42–$175 pricing; IG+FB only; broken footer About link).

Deployable artifacts produced with this report

• /root/cro-reports/shopdeareden/20260826-102324/aeo-deploy/pricing.md — corrected structured pricing file (upload to site root).
• /root/cro-reports/shopdeareden/20260826-102324/aeo-deploy/llms.txt — corrected llmstxt index (upload to site root, replacing the agents.md duplicate).

Confidence: technical checks (robots, files, schema) are High — directly observed via curl. Visibility probes are High for the specific queries tested; broader query coverage would require a paid AI-visibility tool (Otterly/Peec) or larger manual sampling.