simmons-simmons.com — AI Site Grade

Analysis

I have enough data now. Let me compile the audit.

The key findings are:

1. OAI-SearchBot blind spot: The site uses prerender.io (detected via window.prerenderReady = false;). GPTBot, ClaudeBot, Google-Extended, ChatGPT-User, PerplexityBot, Perplexity-User, anthropic-ai, Applebot-Extended, Bytespider all get the full prerendered page (~178KB). But OAI-SearchBot gets the JS shell (1720 bytes) — same as a regular browser. This means ChatGPT's web search feature and GPT-4o's browsing see zero content.

2. No llms.txt: The /llms.txt URL returns the Vue.js app shell (HTML), not a real llms.txt file.

3. No AI-specific robots.txt rules: Only a single User-agent: * rule. No explicit GPTBot, ClaudeBot, Google-Extended, or PerplexityBot directives. The disallow rules block */insights/?, */news/?, */people/?, */profile$, and */search — but these use wildcard patterns that may not match the actual URL structure (which uses paths like /en/publications/..., /en/people/..., /en/about-us/news/...).

4. Schema is thin: The homepage only has a VideoObject schema (for a Wistia embed) and a Corporation schema with basic name/URL/sameAs. No LegalService, Attorney, Organization with detailed practice area schema. No FAQPage, Article, or NewsArticle schema on publications.

5. Sitemap is massive but hidden: The sitemap lives at api.simmons-simmons.com/marketing/sitemap (a subdomain) and contains ~2.5MB of URLs. It's a sitemap index with CDATA-wrapped URLs.

6. Cold knowledge gap: The LLM knows Simmons & Simmons as a London-founded (1896) global law firm strong in financial services, asset management, life sciences. But the site's current positioning ("Together, we deliver smarter solutions", "The Continuum+ Planning, power and people", "STRIDE" AI regulation tracker, "Purpose 2026" video) is completely absent from the model's cold knowledge.

7. No meta descriptions: The homepage and subpages lack meta description tags in the live HTML.

Let me write the audit.

Simmons & Simmons — AI-Visibility Audit

OAI-SearchBot — the bot powering ChatGPT's live web search — receives a 1,720-byte JavaScript shell with zero content, while every other major AI crawler (GPTBot, ClaudeBot, Google-Extended, PerplexityBot, ChatGPT-User) gets a fully prerendered 178KB page. This creates a critical blind spot: any AI query that triggers live retrieval via OpenAI's search pipeline sees nothing but a <noscript> warning.

Crawler Access

The site runs a Vue.js SPA behind a prerender.io service (window.prerenderReady = false). The prerender service correctly serves rendered HTML to GPTBot, ClaudeBot, Google-Extended, PerplexityBot, ChatGPT-User, anthropic-ai, Applebot-Extended, and Bytespider — all returning ~178KB with visible text. OAI-SearchBot is the sole exclusion, receiving the raw JS shell identical to a browser without JavaScript. The robots.txt has a single User-agent: * rule with no AI-bot-specific directives. The disallow patterns (*/insights/?, */news/?, */people/?) use query-string wildcards that likely do not match the actual URL structure (/en/publications/..., /en/people/..., /en/about-us/news/...). No llms.txt exists — the URL returns the Vue app shell HTML instead of a text file.

Content & Schema Posture

The homepage (visible via Wayback snapshot) positions the firm around "The Continuum+ Planning, power and people" and "STRIDE: a new AI-ready digital regulation tracker" — messaging absent from the LLM's cold knowledge. Schema is limited to a VideoObject (Wistia embed) and a bare Corporation type with name, URL, and social links. No LegalService, Attorney, Organization with practice-area schema, no Article or NewsArticle on publications, no FAQPage. The sitemap at api.simmons-simmons.com/marketing/sitemap is a 2.5MB index containing thousands of URLs (publications, people, events, features) with priority: 1.0 on every entry — an unweighted flat structure. No meta description tags appear in the live HTML.

Cold-Knowledge Gap

The LLM describes Simmons & Simmons as a London-founded (1896) firm strong in financial services regulation, fund formation, and life sciences IP, with offices in Hong Kong, Singapore, and Dubai. The site's current strategic messaging — "Purpose 2026", "Continuum+", "STRIDE" AI regulation tracker, the tagline "Together, we deliver smarter solutions" — is entirely absent from the model's prior. The site publishes heavily on AI regulation, ESG, and digital assets (e.g., "AI Omnibus package to amend EU AI Act", "STRIDE" tracker), but the model's knowledge predates these offerings.

External Signals

No indexed external press or Reddit threads surfaced for the current brand messaging. The firm's DNS uses Mimecast for email security and CSC for DNS hosting. The IP (51.140.185.151) resolves to Azure, consistent with the onlineplatform.azurewebsites.net TXT record. The Wayback Machine shows a snapshot from March 2026 with full content, confirming the prerender has been operational for at least several months.

Findings

1. OAI-SearchBot receives empty JavaScript shell High

   OAI-SearchBot, the bot powering ChatGPT's live web search, receives a 1,720-byte JavaScript shell with no content, while GPTBot, ClaudeBot, Google-Extended, PerplexityBot, ChatGPT-User, and others get fully prerendered 178KB pages. This means any AI query triggering live retrieval via OpenAI's search pipeline sees nothing.

   What to change: Add OAI-SearchBot to the prerender.io whitelist so it receives the same prerendered HTML as other AI crawlers.

2. No llms.txt file available Medium

   The /llms.txt URL returns the Vue.js app shell HTML instead of a real llms.txt file, missing an opportunity to guide AI models with structured site information.

   What to change: Create a proper llms.txt file at /llms.txt with a summary of the firm, key pages, and guidance for AI crawlers.

3. No AI-specific robots.txt directives Medium

   The robots.txt contains only a single User-agent: * rule with no explicit directives for GPTBot, ClaudeBot, Google-Extended, or PerplexityBot. The disallow patterns use query-string wildcards that likely do not match the actual URL structure.

   What to change: Add explicit rules for AI bots (e.g., GPTBot, ClaudeBot, Google-Extended) and update disallow patterns to match the actual path structure.

4. Schema markup is minimal and lacks legal-specific types High

   The homepage only has a VideoObject schema and a bare Corporation schema with name, URL, and sameAs. No LegalService, Attorney, Organization with practice-area schema, no Article or NewsArticle on publications, no FAQPage.

   What to change: Add LegalService, Attorney, and Organization schema with practice areas, office locations, and attorney profiles. Add Article or NewsArticle schema to publication pages.

5. Sitemap hosted on subdomain with flat priority Medium

   The sitemap is hosted at api.simmons-simmons.com/marketing/sitemap, a subdomain, and contains a 2.5MB index with thousands of URLs all set to priority 1.0, providing no signal of relative importance.

   What to change: Move the sitemap to the main domain and assign meaningful priority values based on page importance.

6. Current strategic messaging absent from LLM cold knowledge Medium

   The LLM's prior knowledge of Simmons & Simmons lacks the site's current messaging: 'Purpose 2026', 'Continuum+', 'STRIDE' AI regulation tracker, and the tagline 'Together, we deliver smarter solutions'. The model knows the firm as a London-founded global law firm but not its recent AI and digital assets focus.

   What to change: Ensure key strategic pages are well-indexed and include rich schema markup to help AI models associate the firm with its current offerings.

7. Missing meta description tags Low

   The homepage and subpages lack meta description tags in the live HTML, reducing click-through potential in search results and limiting context for AI crawlers.

   What to change: Add unique meta description tags to all key pages.

What's working

• Prerender.io serves full HTML to most AI crawlers — GPTBot, ClaudeBot, Google-Extended, PerplexityBot, ChatGPT-User, and others receive fully prerendered 178KB pages with visible text, ensuring content is accessible to the majority of AI crawlers.
• Sitemap contains thousands of URLs covering publications, people, and events — The sitemap at api.simmons-simmons.com/marketing/sitemap is a 2.5MB index with thousands of URLs, ensuring broad coverage of the site's content for search engines.
• Wayback Machine snapshot confirms prerender has been operational for months — A Wayback Machine snapshot from March 2026 shows full content, indicating the prerender service has been reliably serving rendered pages for at least several months.