bloomberg.com vs rakuten-sec.co.jp — AI Visibility Gap

📊 Evidence Dashboard

Evidence Score

63/100

Composite (retrieval 50% + hallucination 30% + bloat 10% + rank 10%)

Retrieval Mismatch

67%

2/3 queries returned a competitor instead

Hallucination Rate

75%

2/8 fields extracted correctly

Token Bloat

11x

HTML tokens per useful content token

🔍 Shadow RAG Probe Results

Each query was run against our FAISS-indexed corpus of 100,000+ domains. The table shows who the retrieval pipeline cited as the top answer.

Query	Type	Top Retrieved	ACRI	Result
What is Bloomberg?	entity	bloomberg.com	15	✓ Match
What features does Bloomberg offer?	product	rakuten-sec.co.jp	73	⚠ Mismatch
best online platform for enterprise	comparison	rakuten-sec.co.jp	73	⚠ Mismatch

🔧 Remediation Checklist for bloomberg.com

Prioritized technical fixes to reclaim AI citation share. Total estimated effort: 1–3 engineering sprints.

1

Reduce 11x token bloat to under 3x

Your page sends 11 tokens of HTML for every token of useful content. AI crawlers waste context window on navigation, scripts, and boilerplate. Serve a clean semantic layer.

Impact: high
```

<link rel="alternate" type="application/ld+json" href="/structured-data.json">

```

2

Add comprehensive JSON-LD structured data

Only 0 schema types detected. Add Organization, Product/Service, FAQ, and Breadcrumb schemas to give AI extractors ground truth.

Impact: high

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "bloomberg.com",
  "url": "https://bloomberg.com",
  "description": "..."
}
</script>

3

Unblock GPTBot, ClaudeBot in robots.txt

These AI crawlers are currently blocked from reading your content. Domains that allow these bots tend to see higher AI extractability scores.

Impact: critical
```
# Add to robots.txt:
User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /
```
4

Add descriptive meta description

No meta description found. AI extractors use this as a primary signal for page summary. Without it, the model must derive context from noisy body HTML.

Impact: medium
5

Improve semantic HTML structure

Low information density. Use proper headings (H1-H3), ordered lists, definition lists, and table markup so AI can parse your content hierarchically.

Impact: medium

AI Visibility Gap Identified

Assessment: AI Visibility Gap