fabricsocial.com — AI Site Grade

Analysis

Fabric Social: A social-first agency that AI models mistake for a fabric retailer

The cold LLM knowledge describes fabricsocial.com as an online fabric retailer selling Tula Pink and Kaffe Fassett quilting supplies — a complete category error. The site is actually a London-based social creative agency (founded 2021, ~51-200 employees) that was acquired by Publicis Groupe in 2026, with clients including Currys, Sky, Subway, Autotrader, and Morrisons.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended — receive a 200 status with full HTML content identical to browser delivery. No UA-based blocking exists. The site runs on Cloudflare (104.21.76.102 / 172.67.193.11) behind a custom Laravel 12 / Acorn 5.0.2 stack served via NIMHOST. The robots.txt is minimal: a single User-agent: * rule disallowing only /wp-admin/ with no AI-bot-specific directives whatsoever. No llms.txt exists (404). The sitemap contains exactly 3 URLs — homepage, privacy policy, and terms — meaning the site has no discoverable blog, case studies, work pages, or team pages indexed for crawlers.

Cold-Knowledge Gap

The LLM prior describes "high-quality designer-inspired fabrics," "pre-cut fabric bundles," "fat quarters," and "charm packs" from brands like Tula Pink and Ruby Star Society. The actual site is a social media agency whose homepage headline reads "Where Content Meets Culture" and "We transform brands on social into social brands." The model hallucinated an entirely different business category. The gap is total: the LLM knows nothing about the agency's actual positioning, its Publicis Groupe acquisition, its client roster, or its award wins (Global Social Media Awards 2025 Best Large Agency, Cannes Lions winner).

Schema Posture

The site uses JSON-LD schema via The SEO Framework, declaring WebSite, WebPage, BreadcrumbList, and Organization types. The Organization schema correctly lists name, URL, and sameAs links to Instagram and LinkedIn. However, no Service or CreativeWork schema describes the agency's actual offerings (strategy, community, creators, production, distribution). No FAQPage schema exists despite the site having no FAQ content at all. The schema is structurally correct but semantically thin — it identifies the entity as an organization without communicating what it does.

External Signals

LinkedIn shows 17,583 followers and describes the agency as "Best Agency, Global Social Media Awards. Cannes Lions Winner. Acquired by Publicis Groupe in 2026." The LinkedIn page lists 96 employees and active client case studies (Haven pool stunt: 1.6M views in 24 hours; Autotrader: 1.5M views across TikTok; Sky ambassador program). The privacy policy lists a different address (167-169 Great Portland Street) than the footer (60 Worship Street), suggesting a recent office move not reflected consistently. The Campaign article about the Publicis acquisition (linked from the homepage notification bar) returns a 403 to automated fetches, making it invisible to crawlers.

Content Surface Area

The entire discoverable site is 3 pages — homepage, privacy policy, terms. No blog, no case studies page, no work/portfolio page, no team page, no services detail page. The homepage contains only 139 words of visible text. The notification bar ("Breaking news. Fabric Social joins Publicis Groupe") is the most newsworthy item on the site and is rendered as an absolute-positioned element linking to a paywalled/403 Campaign article. The footer copyright reads "© 2026" — a forward-dated year that signals the site may not have been updated recently despite the acquisition being current news.

Findings

1. LLMs hallucinate Fabric Social as a fabric retailer instead of a social agency High

   Cold LLM knowledge describes fabricsocial.com as an online fabric retailer selling quilting supplies, a complete category error. The actual site is a London-based social creative agency acquired by Publicis Groupe in 2026. The gap is total: the LLM knows nothing about the agency's actual positioning, client roster, or awards.

   What to change: Expand the site with substantive content describing the agency's services, clients, case studies, and awards. Add structured data (Service, CreativeWork) to help LLMs correctly categorize the business.

2. Entire discoverable site is only 3 pages with 139 words on homepage High

   The sitemap contains exactly 3 URLs: homepage, privacy policy, and terms. The homepage has only 139 words of visible text. No blog, case studies, work portfolio, team page, or services detail page exists. This extreme thinness gives crawlers almost nothing to index.

   What to change: Create and index pages for services, case studies, team, and blog. Add these to the sitemap. Ensure each page has substantial, unique content.

3. No llms.txt file exists for AI crawlers Medium

   The site returns a 404 for /llms.txt, missing an opportunity to provide LLMs with a curated summary of the agency's identity, offerings, and key pages.

   What to change: Create an llms.txt file that describes the agency, lists key pages, and provides a brief overview to guide LLM understanding.

4. Schema lacks Service or CreativeWork types to describe agency offerings Medium

   The site uses JSON-LD for WebSite, WebPage, BreadcrumbList, and Organization, but no Service or CreativeWork schema describes the agency's actual services (strategy, community, creators, production, distribution). The schema identifies the entity as an organization without communicating what it does.

   What to change: Add Service schema for each agency offering and CreativeWork schema for case studies and awards. Include description, provider, and areaServed properties.

5. Acquisition announcement link returns 403 to crawlers Medium

   The homepage notification bar links to a Campaign article about the Publicis Groupe acquisition, but the article returns a 403 to automated fetches, making the most newsworthy item invisible to crawlers.

   What to change: Publish a press release or blog post on the agency's own domain about the acquisition, and link to that instead of the paywalled article.

6. Footer copyright reads © 2026, suggesting outdated site maintenance Low

   The footer displays '© 2026', a forward-dated year that may signal the site has not been updated recently despite the acquisition being current news. This can erode trust with visitors and crawlers.

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

7. Privacy policy and footer list different office addresses Low

   The privacy policy lists 167-169 Great Portland Street while the footer shows 60 Worship Street, suggesting a recent office move not reflected consistently across the site.

   What to change: Reconcile the address across all pages. Update the privacy policy to match the current office location.

8. No external backlinks or mentions found in web search High

   Multiple web searches for 'Fabric Social' agency, its clients, and its Publicis acquisition returned zero results. This indicates extremely low external signal, making it hard for LLMs to discover or verify the agency's existence.

   What to change: Build a backlink profile through PR, guest posting, and partnerships. Ensure the agency is listed on relevant directories and industry publications.

What's working

• All major AI crawlers receive full HTML content without blocking — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others all receive a 200 status with full HTML content. No UA-based blocking exists. The robots.txt is minimal and does not disallow any AI bots.
• JSON-LD Organization schema correctly declares name, URL, and social links — The site uses JSON-LD via The SEO Framework with Organization schema that correctly lists the agency name, URL, and sameAs links to Instagram and LinkedIn. This provides a baseline entity identity.
• LinkedIn page has 17,583 followers and detailed agency description — The LinkedIn page describes the agency as 'Best Agency, Global Social Media Awards. Cannes Lions Winner. Acquired by Publicis Groupe in 2026.' It lists 96 employees and active client case studies, providing a strong external signal.
• Site runs on Cloudflare with good uptime and performance — The site is served via Cloudflare (cf-ray headers present), indicating CDN protection and performance benefits. DNS records show A records pointing to Cloudflare IPs.