【AI Visibility Case Study】Known but Never Recommended — 36/100

Case Overview

Imagine a potential customer asking an AI assistant: "What are some recommended oriental perfume brands?" The brand in this case study doesn't appear anywhere in the answer. Yet if that same customer already knows the brand name and searches for it directly, the AI responds positively. This contradiction is the most striking finding of this entire audit.

Across 16 test queries run on four major AI platforms — Claude, ChatGPT, Gemini, and Perplexity — the brand received 2 positive mentions and 3 ambiguous mentions, all exclusively from direct brand-name queries. For industry-intent queries such as "oriental perfume brand recommendations" or "artisan fragrance brands," the brand achieved zero positive mentions across all 10 attempts. Even more telling, Claude responded to a direct brand query with: "I'm not aware of this as an established luxury perfume brand. It's possible that it's a very niche or emerging brand." The AI isn't completely ignorant of this brand — it simply cannot place the brand within a recognizable category, and that ambiguity causes AI systems to skip it entirely when generating competitive recommendations.

This case study breaks down the technical, content, and competitive factors behind this "known but never recommended" AI visibility gap — and outlines a clear path forward.

Composite Score Breakdown

The overall audit score is 36 out of 100. The three-dimensional scoring reveals where the structural problems originate.

The most urgent bottleneck is the GEO technical score of just 20 out of 100 — only 3 out of 15 checks passed, and Schema structured data is entirely absent. This means that even when an AI crawler successfully visits the company's website, it cannot reliably extract brand identity, product categories, or content positioning. The result: the brand is effectively invisible in industry-level recommendation scenarios, regardless of how good the products actually are.

AI Visibility Testing Results

Test Summary: Out of 16 queries across four platforms, the brand was mentioned 6 times — but every single mention came from direct brand-name queries. Industry-context queries produced zero mentions.

The Most Significant Finding

Claude's response during direct brand-name testing exposed a critical knowledge gap. Rather than confirming the brand as an established fragrance house, Claude openly stated it could not verify the brand's status and speculated it might be a "very niche or emerging brand." This is not total invisibility — it's something more dangerous: a state of fuzzy recognition that causes AI systems to bypass the brand in recommendation contexts to avoid offering unverified information.

What makes this finding even more instructive is a direct comparison: when queried for "recommended oriental perfume brands," Claude's response included Diptyque Taiwan (a limited series) — a non-local brand that appeared on a "local recommendation" list simply because it had localized, structured content available for AI to read. This reveals a core truth about how AI recommendations work: they are not based on brand authenticity or product quality, but on the density and clarity of machine-readable information.

Platform-by-Platform Results

✓ Positive mention    △ Ambiguous mention    ✗ Not mentioned

Key Insight

The 2 positive mentions from brand queries confirm this brand does exist in AI training data. But the complete absence from 10 industry-context queries indicates that AI systems have not mapped this brand into the "oriental perfume" knowledge category. The most likely root cause is the total absence of structured data — the AI knows the name, but has no reliable signals telling it what category the brand belongs to or what customer need it addresses.

Competitive Landscape Analysis

During industry-level queries, the AI platforms consistently recommended the following competitors: P. Seven, Diptyque, Tom Ford, Dior, Maison Margiela, Yves Rocher, Aesop, and Crabtree & Evelyn.

Notably, all 8 brands recommended by AI are international labels. Local oriental fragrance brands are essentially absent from AI recommendation lists — and this represents both a challenge and a significant opportunity. When every AI-recommended competitor is a global heritage brand, any local brand that successfully builds AI-readable knowledge around "oriental aesthetic fragrance" or "locally crafted artisan perfume" stands to become the default recommendation in that sub-category. The problem here is not that the competition is too fierce — it's that the brand has not yet entered the AI competitive arena at all. The space is open; the infrastructure to claim it is simply not in place yet.

GEO Technical Audit

The technical audit is the most critical weakness in this case. Only 3 out of 15 checks passed — a pass rate of just 20%.

The most consequential gap is the complete absence of Schema JSON-LD markup. For AI systems, Schema markup is the primary semantic signal for understanding what a website sells, who the brand is, and what products exist. Without Product Schema, an AI crawler cannot confirm that a specific fragrance is a perfume product. Without Brand Schema, it cannot map the company to any product category. This directly explains why brand-name queries yield ambiguous responses rather than confident, category-specific recommendations.

For more context on GEO technical audit methodology, explore additional industry reports in the Joseph Intelligence Case Study Index.

Website Performance

The composite PageSpeed score of 70 looks passable at first glance — but the underlying data reveals a serious underlying problem.

• Performance Score: 57 / 100
• SEO Score: 82 / 100
• LCP (Largest Contentful Paint): 38,552 milliseconds — approximately 15,420 times above Google's recommended threshold of 2,500ms

An LCP of 38,552ms is the single most severe performance problem in this audit. LCP measures how quickly a page's primary content becomes visible to users — and to crawlers. When an AI crawler cannot retrieve above-the-fold content within a reasonable timeframe, it is highly likely to abort indexing early, leaving a permanent gap between what the company's website actually contains and what AI systems record. Fragrance brand websites frequently rely on large, high-resolution product images with immersive visual layouts — a design approach that delivers beautifully for human visitors but creates severe rendering bottlenecks for automated crawlers. This site appears to be a textbook example of that trade-off.

Expert Recommendations

1. Bring LCP from 38,552ms Down to Under 2,500ms — Make the Homepage Crawlable

At its current LCP, AI crawlers are almost certainly timing out before they can index the homepage's full content. The likely causes include uncompressed high-resolution product images, render-blocking JavaScript, and unoptimized CSS delivery. Recommended actions: convert product imagery to WebP format with lazy loading enabled, preload the critical above-the-fold image resource, and defer non-essential JavaScript to asynchronous loading. Reaching a Performance score of 75 or above is a prerequisite for any other GEO optimization to take effect — it's the foundation everything else depends on.

2. Build AI-Readable Fragrance Knowledge Pages for Hero Products

When fragrance buyers consult AI, they're not just asking for brand names — they're asking questions like "what occasions suit a kashmir ambergris scent?" or "what's the difference between oriental and woody fragrances?" The company's website currently lacks structured content that answers these questions, which explains the zero-mention rate across all 10 industry queries. For each core product, the recommendation is to create dedicated depth pages covering fragrance family description, usage occasions, and cultural context — then mark up each page with both Product Schema and FAQ Schema so AI systems can map that content directly to relevant buyer queries.

3. Claim "Local Oriental Aesthetic" as a Differentiation Anchor in AI Knowledge Graphs

All 8 AI-recommended competitors in this audit are global brands. Their interpretation of "oriental" is, by definition, an outsider's perspective. A brand rooted in local culture has a genuine, defensible advantage in articulating what oriental fragrance means from the inside — its connection to local lifestyle, seasonal rhythms, and aesthetic philosophy. The strategic recommendation is to produce definitional content around precise category terms that international brands cannot authentically own. Being the first local brand to establish that content footprint means becoming the default AI recommendation in that sub-category — a position that compounds in value as AI search continues to grow.

AI Search Trends in the 香水香氛產業、東方香水品牌 Industry

The fragrance industry has a fundamentally different AI search dynamic compared to most e-commerce categories. Perfume is a sensory product — but AI works exclusively with text and structured data. This tension creates unique risks for fragrance brands that haven't deliberately built machine-readable content strategies.

Two Buyer Profiles and How They Search

Fragrance buyers typically fall into two search patterns. The first is the occasion-driven buyer — they know they need a fragrance for a specific purpose (autumn gifting, a formal event, a luxury self-purchase) but haven't chosen a brand. The second is the aesthetic explorer — they have a general familiarity with fragrance families such as oriental, woody, or musky, and are looking for brands that match a particular aesthetic sensibility. Both types consult AI before purchasing, but their query structures differ significantly. A brand with no structured content has essentially no pathway to reach either group through AI search.

The Questions Buyers Most Likely Ask AI

• "What oriental eau de parfum works well in humid climates?"
• "What fragrance family is Kashmir ambergris? Who is it for?"
• "Are there any independent fragrance brands with a genuine East Asian aesthetic identity?"

Each of these questions requires AI to have fragrance category knowledge, cultural context, and a clear brand-to-category mapping. All three requirements are met through Product Schema, FAQ Schema, and substantive depth content — precisely the elements absent from the company's website in this audit.

Why Fragrance Websites Are Especially Vulnerable to AI Visibility Loss

Fragrance brand websites are typically designed around visual immersion — large imagery, animated transitions, and cinematic layouts. This creates an exceptional user experience for human visitors while simultaneously making the site nearly unreadable for AI crawlers. When a page's primary content is delivered through images rather than structured text, AI systems cannot extract brand positioning, product characteristics, or fragrance descriptions. The extreme LCP of 38,552ms in this audit is a direct consequence of this "design-first, semantics-absent" architecture — a pattern common across luxury fragrance brands globally.

How International Brand Dominance Shapes AI Recommendations

When an AI system processes a query like "oriental perfume recommendation," it draws on brands with the highest knowledge density in its training data. Tom Ford, Diptyque, and Maison Margiela have accumulated decades of English-language reviews, fragrance database entries, editorial coverage, and structured product information. This accumulated data is what drives AI recommendations — not brand prestige alone. The competitive disadvantage for independent local brands is not product quality; it's the sheer volume of AI-readable information that global brands have built up over time.

The Untapped AI Visibility Opportunity for Local Oriental Fragrance

Across major English-language AI platforms, content about locally produced oriental fragrance brands is remarkably sparse. Even a query like "Taiwan oriental perfume brand" returns almost no local results. For any brand whose core positioning centers on oriental aesthetics and local cultural identity, this represents a genuine blue ocean in AI visibility — a category where no competitor has yet established dominance in the AI knowledge graph. Capturing that space requires bilingual, structured, AI-citable content. To assess where your own brand stands right now, use the free AI Visibility Self-Check tool.

Wondering How Your Brand Appears in AI Search?

If your brand faces a similar "recognized but never recommended" problem — or if you simply don't know how you're performing across AI platforms — here are two ways to find out:

• Run the free AI Visibility Self-Check and get an initial assessment in under 3 minutes.
• Or schedule a free results consultation with a Joseph Intelligence advisor for a deep-dive analysis tailored to your industry and competitive landscape.

Disclaimer

This article is based on anonymized real audit data. All identifiable company information has been removed. AI platform responses are non-deterministic and may vary over time. Technical audit and performance scores are point-in-time snapshots.