granularit.com — AI Site Grade

Analysis

AI crawlers see this site fully — but find almost nothing to work with

Every major AI crawler (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended) receives a 200 with full HTML content from Cloudflare. No UA-based blocks, no JS shell, no 403s. The technical access posture is wide open. Yet the cold-knowledge gap is absolute: a frontier LLM queried on "Granular IT" returns zero facts — cannot confirm services, location, or even that the company exists.

Crawler Access

The robots.txt at granularit.com/robots.txt contains zero AI-bot directives. No User-agent: GPTBot, no User-agent: ClaudeBot, no User-agent: Google-Extended. The wildcard rule (Allow: /) governs all crawlers, with only /admin/ and /_private/ disallowed. The llms.txt endpoint returns a full HTML 404 page (200 status, "Page Not Found" title, 17KB of rendered Remix framework boilerplate) — not a 404 status code, but a soft 404 page. This means any AI engine looking for an llms.txt content map gets a broken experience. The site runs on Cloudflare with Remix (React SSR framework), so content is server-rendered — no JS-rendering risk for crawlers.

Cold-Knowledge Gap

A frontier LLM queried cold on "Granular IT" reports: *"I do not have specific, detailed knowledge about a company named 'Granular IT'... It is not a widely recognized or frequently referenced entity."* The model confuses the name with Granular (farm software by Corteva) and can only speculate. This is a complete knowledge vacuum despite the site having 40+ indexed pages, detailed service descriptions, transparent pricing, and case studies. The site's actual content — a dual-location MSP (Nairobi + London) offering managed IT, CCTV, access control, fire alarms, AV, fiber, and solar — is entirely absent from the model's prior.

Schema Posture

The homepage carries two LocalBusiness JSON-LD blocks — one for Kenya (@id: granularit.com/#nairobi) and one for London (@id: granularit.com/#london) — with correct addresses, geo coordinates, phone numbers, and service types. The FAQ page has a FAQPage schema with 15+ questions covering coverage areas, SLAs, pricing, and security. The case study page uses Article schema with datePublished: 2026-05-30 (a future date). However, the /projects listing page has a broken BreadcrumbList with duplicate "Home" entries at positions 1 and 2. The /pricing and /contact pages are missing canonical tags. The /sustainability page has zero schema markup at all.

External Signals

Web searches for "granularit.com", "Granular IT Nairobi", and "Granular IT reviews" return zero indexed results on DuckDuckGo. No Reddit threads, no press mentions, no review sites, no directory listings surfaced. The Wayback Machine has no snapshots of the domain — it has never been archived. DNS records show Google Workspace (MX), Brevo, Infusionsoft, and Mailchimp (SPF include) — suggesting active email marketing and CRM tooling — but none of this translates to external visibility. The site links to four.africa as a sister agency for custom development, but that domain is similarly obscure. The brand exists in a near-total off-domain silence that reinforces the cold-knowledge vacuum.

Surprising Details

The copyright footer reads "© 2026 Granular IT" — two years ahead of the current date. The server response headers include x-render-origin-server and rndr-id, indicating the site uses a server-side rendering service (likely Remix on a Node.js backend behind Cloudflare). The /projects/server-room-reset case study is impressively detailed (6-step method, tech stack, before/after narrative) and carries proper Article schema — but its datePublished is set to May 30, 2026, which is a future date that may confuse temporal reasoning in AI parsers. The /services/managed-it-services/it-support-london page explicitly guarantees "zero cross-border data sharing" between UK and Kenya operations, a strong GDPR selling point that no AI model currently knows about.

Findings

1. Complete cold-knowledge vacuum for Granular IT High

   A frontier LLM queried on 'Granular IT' returns zero facts, cannot confirm services, location, or existence, and confuses the brand with unrelated software. The site's detailed content is entirely absent from the model's prior.

   What to change: Build external citations through directory listings, press releases, guest posts, and review platforms to create off-domain signals that AI models can index.

2. Zero external signals from web searches High

   Web searches for 'granularit.com', 'Granular IT Nairobi', and 'Granular IT reviews' return zero indexed results on DuckDuckGo. No Reddit threads, press mentions, review sites, or directory listings exist. The Wayback Machine has no snapshots of the domain.

   What to change: Submit the site to business directories (Google Business Profile, Bing Places, Yelp), publish case studies on external platforms, and engage in industry forums to generate backlinks and mentions.

3. llms.txt endpoint returns a soft 404 page High

   The /llms.txt endpoint returns a full HTML 404 page (200 status, 'Page Not Found' title, 17KB of Remix boilerplate) instead of a proper 404 status or a valid llms.txt file. AI engines looking for a content map get a broken experience.

   What to change: Create a valid llms.txt file at /llms.txt listing key pages and summaries, or configure the server to return a 404 status code for missing resources.

4. Case study schema uses future dates Medium

   The /projects/server-room-reset case study has Article schema with datePublished set to May 30, 2026, which is a future date. This may confuse temporal reasoning in AI parsers and reduce credibility.

   What to change: Update the datePublished field to the actual publication date or use a valid past date.

5. Broken BreadcrumbList with duplicate entries Medium

   The /projects listing page has a BreadcrumbList schema with duplicate 'Home' entries at positions 1 and 2, which violates schema best practices and may confuse parsers.

   What to change: Fix the BreadcrumbList to have unique, sequential entries without duplicates.

6. Missing canonical tags on pricing and contact pages Medium

   The /pricing and /contact pages lack canonical tags, which can lead to duplicate content issues and dilute SEO signals.

   What to change: Add self-referencing canonical tags to all pages, especially /pricing and /contact.

7. Sustainability page has zero schema markup Medium

   The /sustainability page, which describes sustainable IT solutions, has no structured data at all, missing an opportunity to be featured in rich results.

   What to change: Add appropriate schema markup (e.g., Article or WebPage with relevant properties) to the sustainability page.

8. Copyright footer shows future year 2026 Low

   The site footer displays '© 2026 Granular IT', which is two years ahead of the current date. This may appear unprofessional and could confuse temporal signals.

   What to change: Update the copyright year to the current year or use dynamic year rendering.

9. No AI bot directives in robots.txt Low

   The robots.txt file contains no User-agent lines for any AI crawler (GPTBot, ClaudeBot, etc.), relying solely on the wildcard rule. While this allows access, it misses the opportunity to signal crawl preferences or prioritize certain paths.

   What to change: Add explicit User-agent directives for major AI crawlers to control crawl behavior and signal importance.

What's working

• All major AI crawlers receive full HTML content — Every tested AI bot (GPTBot, ClaudeBot, PerplexityBot, etc.) gets a 200 response with full server-rendered HTML from Cloudflare. No UA-based blocks, JS shells, or 403 errors exist.
• Homepage has correct LocalBusiness schema for both locations — The homepage carries two LocalBusiness JSON-LD blocks with accurate addresses, geo coordinates, phone numbers, and service types for Nairobi and London.
• FAQ page has FAQPage schema with 15+ questions — The /faq page includes FAQPage structured data covering coverage areas, SLAs, pricing, and security, which can enable rich results in search.
• Case study page has detailed Article schema — The /projects/server-room-reset case study uses Article schema with a detailed narrative, tech stack, and before/after description, providing rich content for AI parsers.
• Server-rendered content via Remix avoids JS dependency — The site uses Remix (React SSR) behind Cloudflare, ensuring all content is server-rendered and fully visible to crawlers without JavaScript execution.
• London page highlights zero cross-border data sharing — The /services/managed-it-services/it-support-london page explicitly guarantees zero cross-border data sharing between UK and Kenya, a strong GDPR compliance selling point.