webstarsltd.com — AI Site Grade

Analysis

The site runs a modern JS-heavy SPA on Netlify where 7 of 14 discovered pages return under 10 words of visible text to plain HTTP GET — including the products page, the contact page, and the entire insights listing page.

Crawler Access

All major AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended) receive identical 200 responses as a browser — no UA-based blocking exists. The site is hosted on Netlify with strong security headers (HSTS, CSP, X-Frame-Options). The robots.txt is a bare allow-all (User-agent: * Disallow:) with no AI-bot-specific directives. The llms.txt endpoint returns HTTP 500 — not a 404, but a server error that prevents any AI-friendly content map from being served.

JS-Rendering Crisis

The homepage yields only 75 words of visible text. The /insights/ listing page, /products/, /contact, and /webstars-whispers/ each return fewer than 10 words — just a heading and the word "Loading." These are JavaScript shells. AI crawlers that do not execute JS (the vast majority) see essentially blank pages. The sitemap (sitemap-index.xml) lists 5 URLs, but the sitemap-0.xml sub-sitemap was not directly accessible. The blog posts and case studies that do render server-side (e.g. /insights/the-power-of-data-visualisation, /projects/kingston-technology) are well-written, substantive content — but they are buried behind a JS-dependent navigation layer that crawlers cannot traverse.

Cold-Knowledge Gap

A frontier LLM queried cold about webstarsltd.com returned: "I do not have specific, verifiable information about webstarsltd.com." The model knows nothing about this agency — not its name, its London location, its 23-year history, its ABM expertise, or its client roster (Kingston Technology, Proskauer, William Hill, Sumitomo Mitsui). The site claims to have operated for 23 years, yet has effectively zero external footprint: DuckDuckGo searches for the brand name, its directors, and its case study clients returned zero results. No reviews, no press mentions, no Reddit threads, no external backlinks were found.

Schema Posture

Zero JSON-LD schema exists on any page examined — not on the homepage, not on blog articles, not on case studies. No Organization, LocalBusiness, Article, FAQPage, or BreadcrumbList markup is present. The blog posts use proper og:type: article Open Graph tags and have clean <h1>/<h2> heading hierarchies, but the absence of structured data means AI engines cannot extract entity relationships, authorship, or service offerings programmatically.

Content Fragmentation

The blog content is genuinely useful (ABM methodology, data visualisation, ROMI, zero-party data) and was updated as recently as February 2025 (PPC vs. SEO article). However, the ABM Stage 1 article is dated October 2019 and references outdated tools (Eloqua, Pardot). The site links to a "Complete Guide" PDF (/90-second-guide-updated.pdf) from nearly every page, but the PDF was not fetched. The case studies are templated with consistent headings (Objectives, Features, Services, Results) — a format AI engines could parse well — but without schema markup, the structure is invisible to them.

Findings

1. Seven pages render as empty JavaScript shells to plain HTTP GET High

   The homepage yields only 75 words of visible text; the insights listing, products, contact, and webstars-whispers pages each return fewer than 10 words — just a heading and 'Loading.' AI crawlers that do not execute JavaScript see essentially blank pages.

   What to change: Implement server-side rendering (SSR) or static pre-rendering for all pages, especially the insights listing, products, and contact pages, so that AI crawlers receive meaningful HTML content without requiring JavaScript execution.

2. llms.txt endpoint returns HTTP 500 error Medium

   The llms.txt endpoint returns a server error (HTTP 500) instead of a 404 or a valid file, preventing AI-friendly content maps from being served.

   What to change: Fix the server error on the llms.txt endpoint and serve a valid llms.txt file listing key URLs and content summaries for AI crawlers.

3. No JSON-LD structured data on any page High

   Zero JSON-LD schema exists on any examined page — no Organization, LocalBusiness, Article, FAQPage, or BreadcrumbList markup. AI engines cannot extract entity relationships, authorship, or service offerings programmatically.

   What to change: Add JSON-LD structured data to all pages: Organization and LocalBusiness schema on the homepage, Article schema on blog posts, and BreadcrumbList on all pages.

4. Near-zero external visibility: no search results, reviews, or backlinks High

   DuckDuckGo searches for the brand name, directors, and case study clients returned zero results. No reviews, press mentions, Reddit threads, or external backlinks were found. A frontier LLM queried cold about the domain returned 'I do not have specific, verifiable information about webstarsltd.com.'

   What to change: Build external signals through PR, guest posting, directory listings, and client testimonials. Ensure the site is indexed by Google and other search engines.

5. Navigation layer requires JavaScript, blocking crawler traversal High

   The site's navigation is JavaScript-dependent, so AI crawlers that do not execute JS cannot discover internal pages beyond the homepage. Substantive blog posts and case studies are buried behind this layer.

   What to change: Add static HTML links or a sitemap with all internal pages to ensure crawlers can discover content without JavaScript.

6. Sub-sitemap not directly accessible Medium

   The sitemap index lists 5 URLs, but the sub-sitemap (sitemap-0.xml) was not directly accessible, potentially hiding pages from crawlers.

   What to change: Ensure all sub-sitemaps are publicly accessible and listed correctly in the sitemap index.

7. ABM Stage 1 article references outdated tools Low

   The ABM Stage 1 article (dated October 2019) references Eloqua and Pardot, which may appear stale to AI engines and readers.

   What to change: Update the article to reference current tools and practices, or add a note that the methodology remains relevant.

What's working

• All major AI crawlers allowed with identical 200 responses — The site does not block any AI crawler via robots.txt or UA-based rules, and all tested bots receive the same 200 response as a browser.
• Blog posts and case studies contain well-written, substantive content — Pages like the data visualisation article (2033 words) and the Kingston Technology case study (271 words) provide detailed, useful information that AI engines could leverage if accessible.
• Case studies use consistent templated headings — Case studies follow a consistent structure with headings like Objectives, Features, Services, and Results, which AI engines could parse well if schema were added.
• Blog posts use proper Open Graph tags and heading hierarchy — Blog articles include og:type: article and clean <h1>/<h2> heading structures, aiding social sharing and basic content extraction.
• Strong security headers (HSTS, CSP, X-Frame-Options) in place — The site serves HSTS, CSP, and X-Frame-Options headers, providing good security posture.