futureshg.co.uk — AI Site Grade

Analysis

Futures Housing Group — AI-Visibility Audit

The site claims G1/V1 (the highest regulatory grade) on its About page, but the cold LLM knowledge base remembers G1/V2 — a material downgrade that the site itself never acknowledges, creating a direct contradiction an AI engine would surface if it cross-referenced both sources.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended — receive a full 200 response with identical content to a browser visit. No UA-based blocking, no WAF challenge, no JS shell. The robots.txt is a bare-bones catch-all (User-agent: *) blocking only /app_plugins/ and /umbraco/ with no AI-specific directives. The llms.txt returns 404. The site runs on a single Azure IP (51.104.28.80) with no CDN layer, served via Umbraco CMS. All 328 sitemap URLs are crawlable.

Cold-Knowledge Gap

The LLM prior describes Futures Housing Group as managing over 8,000 homes and rated G1/V2 by the Regulator of Social Housing. The actual site states more than 10,700 homes and claims G1/V1 — the highest governance and viability grade. The site's own performance page shows a 75.5% overall customer satisfaction score and a 55.9% satisfaction with antisocial behaviour handling, data the cold model does not reference. The model also recalls a commercial subsidiary called "Futures Property Services" for repairs and maintenance, but the site names its commercial arm Futures Living Ltd (development) and mentions Beep Assist (independent living tech) and Access Training — none of which appear in the cold knowledge.

Schema Posture

Zero JSON-LD structured data exists on any page examined — homepage, About, Find a Home, Performance, Governance, or Newsroom. No Organization, LocalBusiness, FAQPage, WebSite, or Service schema is present. The site contains FAQ-style content (the Help Hub, the repairs Q&A) and tabular performance data (Tenant Satisfaction Measures with percentages) that would be strong candidates for FAQPage and structured data markup, but none is implemented. The homepage uses multiple H1 tags (9 instances), which is non-standard and may confuse content extraction.

External Signals

The domain has near-zero external web footprint. Searches for the brand name, regulatory judgements, news coverage, and reviews return no indexed results from DuckDuckGo. The site has no press mentions, no Reddit threads, no review sites referencing it in search results. The only external signals are social media links (Facebook, LinkedIn, YouTube) and partner links (Beep Assist, Atem, Investors in People). This absence of off-domain citations means AI engines have almost no corroborating third-party data to validate or enrich the brand's own claims — making the G1/V1 vs G1/V2 discrepancy particularly risky for retrieval-augmented generation contexts.

Findings

1. Regulatory grade G1/V1 claimed on site contradicts LLM knowledge of G1/V2 High

   The site's About page claims G1/V1 (highest governance and viability grade), but the cold LLM knowledge base recalls G1/V2, a material downgrade. The site never acknowledges the discrepancy, creating a direct contradiction an AI engine would surface if cross-referenced.

   What to change: Update the site to reflect the correct current regulatory grade and add a note explaining any changes, or ensure the LLM knowledge is corrected through external signals.

2. Zero JSON-LD structured data on any page High

   No JSON-LD structured data exists on the homepage, About, Find a Home, Performance, Governance, or Newsroom pages. No Organization, LocalBusiness, FAQPage, WebSite, or Service schema is present, missing opportunities for rich AI extraction.

   What to change: Add JSON-LD structured data for Organization, WebSite, FAQPage (for Help Hub and repairs Q&A), and Service schema for key offerings.

3. Cold LLM knowledge lacks current site data on home count, satisfaction scores, and subsidiaries High

   The LLM prior states 'over 8,000 homes' and G1/V2, while the site says 'more than 10,700 homes' and G1/V1. The site's performance page shows 75.5% customer satisfaction and 55.9% satisfaction with antisocial behaviour handling, none of which appear in cold knowledge. The model recalls 'Futures Property Services' but the site names 'Futures Living Ltd', 'Beep Assist', and 'Access Training'.

   What to change: Publish an llms.txt file and ensure key facts (home count, regulatory grade, satisfaction scores, subsidiaries) are consistently presented and externally cited to update LLM knowledge.

4. Near-zero external web footprint with no indexed news, reviews, or regulatory mentions High

   Searches for the brand name, regulatory judgements, news coverage, and reviews return no indexed results from DuckDuckGo. The domain has no press mentions, Reddit threads, or review site references. This absence of third-party citations means AI engines have little corroborating data to validate the site's claims.

   What to change: Build external signals through PR, press releases, listings on review sites, and participation in industry forums to create a web footprint that AI engines can index.

5. llms.txt file returns 404 Medium

   The llms.txt file, which AI engines use to discover key content, returns a 404 error. This prevents AI crawlers from efficiently finding the site's most important pages and data.

   What to change: Create an llms.txt file listing key pages (About, Performance, Governance, Find a Home, Newsroom) and key facts (home count, regulatory grade, satisfaction scores).

6. Homepage uses multiple H1 tags (9 instances) Low

   The homepage contains 9 H1 tags, which is non-standard and may confuse content extraction by AI crawlers and search engines. Typically, a page should have a single H1.

   What to change: Reduce the homepage to a single H1 tag and use H2-H6 for subheadings.

7. FAQ-style content lacks FAQPage schema Medium

   The site contains FAQ-style content (Help Hub, repairs Q&A) and tabular performance data (Tenant Satisfaction Measures) that would benefit from FAQPage and structured data markup, but none is implemented.

   What to change: Add FAQPage schema to FAQ sections and use structured data for performance metrics (e.g., Dataset or StatisticalData schema).

8. Missing Organization schema on homepage and About page Medium

   The site does not include Organization schema, which would help AI engines understand the entity's name, logo, contact info, and social profiles.

   What to change: Add Organization JSON-LD schema with name, logo, URL, sameAs (social links), and contact info.

9. Missing WebSite schema Low

   The site does not include WebSite schema, which can provide search engines with site name and search action information.

   What to change: Add WebSite schema with site name and potential search action.

10. Missing Service schema for key offerings Medium

    The site offers services like 'Find a Home', repairs, and independent living tech (Beep Assist), but no Service schema is used to describe these offerings.

    What to change: Add Service schema for key offerings such as property search, repairs, and Beep Assist.

11. Missing LocalBusiness schema for physical offices Low

    The site likely has physical offices, but no LocalBusiness schema is present to provide address and contact information.

    What to change: Add LocalBusiness schema for each office location with address, phone, and opening hours.

12. Missing NewsArticle schema on newsroom pages Low

    The newsroom page contains news articles but lacks NewsArticle schema, which would help AI engines surface news content.

    What to change: Add NewsArticle schema to individual news articles with headline, datePublished, and author.

What's working

• All major AI crawlers receive full 200 response with identical content — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, and Applebot-Extended all receive a full 200 response with identical content to a browser visit. No UA-based blocking, WAF challenge, or JS shell.
• Sitemap available with 328 URLs, all crawlable — The sitemap contains 328 URLs and is accessible, ensuring AI crawlers can discover all pages.
• robots.txt allows AI crawlers with no AI-specific blocks — The robots.txt file has a catch-all rule blocking only /app_plugins/ and /umbraco/, with no AI-specific directives, so AI crawlers are not blocked from any meaningful content.
• Performance page provides detailed Tenant Satisfaction Measures — The performance page includes specific satisfaction scores (75.5% overall, 55.9% for antisocial behaviour handling) and other metrics that are valuable for AI extraction, though currently not structured.
• Newsroom page exists with recent articles — The newsroom page contains news articles, providing a source of fresh content that AI engines can index.
• Social media links to Facebook, LinkedIn, YouTube present — The site includes links to its social media profiles, which can help AI engines verify the brand's online presence.
• Clear site structure with key pages (About, Find a Home, Performance, Governance, Newsroom) — The site has a logical structure with dedicated pages for important information, making it easier for AI crawlers to navigate and extract content.
• No JavaScript shell or client-side rendering issues — All pages render server-side HTML, so AI crawlers can access content without executing JavaScript.