gic.com.sg — AI Site Grade

Analysis

The bare domain gic.com.sg has no DNS A/AAAA records — the entire web presence lives at www.gic.com.sg, creating a split-domain identity that AI crawlers hitting the canonical brand domain get nothing from.

Crawler Access

Every major AI bot — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, anthropic-ai — receives a full 200 response with identical HTML content (110KB) from www.gic.com.sg. No UA-based blocking, no Cloudflare challenge, no JS shell. The site runs on nginx behind AWS CloudFront with aggressive security headers (HSTS, CSP, X-Frame-Options). The robots.txt is a bare Yoast default (User-agent: * Disallow:) with no AI-bot-specific rules whatsoever. /llms.txt returns a 404 WordPress page. The sitemap index is healthy with 20 sub-sitemaps covering ~200+ URLs.

Cold-Knowledge Gap

The LLM prior describes GIC as managing "over $700 billion in assets" with a "20-year annualized return at 4.6% as of March 2023" and notable holdings in Citigroup, UBS, and Alibaba. The actual site reports a 3.8% 20-year real rate of return for the year ended 31 March 2025 — a material downward revision. The site deliberately does not disclose AUM, stating that revealing the size poses a national security risk. The cold model cites a figure the site itself refuses to confirm, and cites stale performance data that has since been updated. The model also mentions China property exposure scrutiny — the site makes no mention of this.

Schema Posture

Every page carries the same three Yoast-generated schema types: WebPage, BreadcrumbList, and Organization. The Organization schema is minimal — name, URL, and a generic logo image. No FinancialService, InvestmentFund, GovernmentOrganization, or FAQPage schema types are used anywhere, despite the site having a rich 25-question FAQ page that is structurally perfect for FAQPage markup. No Article schema on ThinkSpace articles. No Report schema on the annual report microsite at report.gic.com.sg.

Content & Answer Signals

The homepage and all subpages repeat the same three H1 headings verbatim, creating heading duplication across every page — a structural SEO weakness. The FAQ page at /who-we-are/faqs/ contains 2,761 words of substantive Q&A content covering performance, governance, investments, and the reserves framework. ThinkSpace publishes long-form thought-leadership articles (1,000+ words each) on AI, macro outlook, and portfolio construction — content that is highly relevant to AI training data but lacks structured data. The Japanese-language subdomain at /jp/ mirrors the English site structure with its own sitemap.

External Signals

GIC maintains active LinkedIn, Instagram, and YouTube presences. The annual report lives on a separate subdomain (report.gic.com.sg) with its own navigation and no schema linkage back to the main site. The careers portal is on gic.careers and careers.gic.com.sg. The Bridge Forum event site is on bridgeforum.io. These fragmented subdomains mean AI crawlers encounter GIC's brand across multiple unlinked digital properties with no canonical consolidation.

Findings

1. Bare domain gic.com.sg has no DNS records High

   The canonical brand domain gic.com.sg has no A/AAAA records, so AI crawlers hitting it receive nothing. All content lives at www.gic.com.sg, creating a split-domain identity.

   What to change: Add a CNAME or A record for gic.com.sg pointing to www.gic.com.sg, or configure a redirect from the bare domain.

2. No /llms.txt file Medium

   The site returns a 404 for /llms.txt, missing an opportunity to guide AI crawlers to key content.

   What to change: Create an /llms.txt file listing important pages like the FAQ, ThinkSpace articles, and annual report.

3. Cold LLM knowledge contains stale and unconfirmed data High

   The LLM prior cites GIC managing over $700 billion in assets and a 4.6% 20-year return, but the site does not disclose AUM and reports a 3.8% return for 2025. The model's information is outdated and unverified.

   What to change: Publish an authoritative data page with current performance figures and AUM rationale, and mark it up with schema to correct LLM knowledge.

4. FAQ page lacks FAQPage structured data High

   The 25-question FAQ page at /who-we-are/faqs/ has no FAQPage schema markup, missing a prime opportunity for rich results and AI training signals.

   What to change: Add FAQPage schema with Question/Answer markup to the FAQ page.

5. No FinancialService or InvestmentFund schema Medium

   Despite being a sovereign wealth fund, the site uses only generic Organization schema. Adding FinancialService or InvestmentFund schema would improve entity recognition.

   What to change: Add FinancialService or InvestmentFund schema to the homepage and about pages.

6. ThinkSpace articles lack Article schema Medium

   Long-form thought-leadership articles on ThinkSpace have no Article or NewsArticle schema, reducing their visibility in AI training and search.

   What to change: Add Article schema to all ThinkSpace articles.

7. Duplicate H1 headings across all pages Medium

   Every page repeats the same three H1 headings, creating structural SEO weakness and confusing crawlers about page hierarchy.

   What to change: Ensure each page has a unique H1 that describes its specific content.

8. Key content on separate subdomains without canonical linkage Medium

   The annual report (report.gic.com.sg), careers (gic.careers), and Bridge Forum (bridgeforum.io) are on separate subdomains with no schema or canonical links back to the main site, diluting brand authority.

   What to change: Add rel=canonical links and Organization schema with sameAs references across all subdomains.

9. Annual report microsite lacks Report schema Medium

   The annual report at report.gic.com.sg has no Report or FinancialReport schema, missing a chance for rich results and AI training signals.

   What to change: Add Report or FinancialReport schema to the annual report pages.

10. Organization schema is minimal Low

    The Organization schema on every page only includes name, URL, and logo. Missing fields like description, sameAs, and foundingDate limit entity understanding.

    What to change: Expand Organization schema with description, sameAs (LinkedIn, YouTube), and foundingDate.

11. Robots.txt has no AI-bot-specific rules Low

    The robots.txt is a bare Yoast default with no directives for AI crawlers, missing an opportunity to guide them to important content.

    What to change: Add specific Allow rules for AI bots to key sections like /thinkspace/ and /who-we-are/faqs/.

What's working

• All major AI bots receive full HTML content — Every tested AI bot (GPTBot, ClaudeBot, PerplexityBot, etc.) gets a 200 response with identical HTML, no blocking or JS shells.
• Sitemap index is well-structured with 20 sub-sitemaps — The sitemap index at /sitemap_index.xml contains 20 sub-sitemaps covering ~200+ URLs, ensuring crawlers can discover all pages.
• FAQ page provides comprehensive Q&A content — The 2,761-word FAQ page covers performance, governance, and reserves, offering valuable training material for AI models.
• ThinkSpace publishes long-form thought-leadership content — Articles like 'The Dawn of AI' provide substantive, AI-relevant content that can be used for training and citations.
• Active LinkedIn, Instagram, and YouTube channels — GIC maintains active social media profiles that can serve as external signals and citation sources for AI models.
• Annual report hosted on dedicated subdomain — The annual report at report.gic.com.sg provides detailed financial data in a dedicated space, though lacking schema linkage.
• Japanese-language version of the site with its own sitemap — The /jp/ subdirectory mirrors the English site and has its own sitemap, expanding international reach.