ricart.com — AI Site Grade

Analysis

Ricart.com's AI crawler access is fully open, yet the site's fragmented multi-domain architecture and stale content create a significant gap between what AI models know cold and what the site actually communicates.

Crawler Access

Every major AI crawler -- GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended, and anthropic-ai -- receives a 200 status with full HTML content on the homepage. No UA-based blocking, no JS shell, no 403. The site runs on nginx behind Varnish CDN with ASP.NET backend (DealerOn platform), and the robots.txt uses a single User-Agent: * rule with a Crawl-delay: 10 but zero AI-bot-specific directives. No AI crawler is explicitly allowed or disallowed. The llms.txt returns a 404 (a branded 404 page, not a bare server error).

Cold-Knowledge Gap

The LLM prior knows Ricart as a "large, family-owned auto dealership group based in Columbus, Ohio, operating since 1953" with a "Mega Mall" concept, "no-haggle" pricing, and mixed reviews citing pressure sales tactics. The site itself, however, never mentions "no-haggle" pricing anywhere on the homepage, about page, or buying-power page. The site emphasizes a "Best Value Guarantee" (refund the difference if a lower price is found) and a "72 Hour Exchange Policy" -- positioning that the cold model does not reference. The model also mentions Mazda as a brand Ricart sells, but the site lists only Buick, Chevrolet, Ford, Genesis, GMC, Hyundai, Kia, Mitsubishi, and Nissan -- no Mazda. The "Mega Mall" concept is mentioned on the homepage (67 acres) but is not a prominent branded term.

Schema Posture

The homepage and all subpages carry AutomotiveBusiness JSON-LD with complete address, geo, telephone, opening hours, department breakdowns (Sales/Service/Parts), and sameAs links to X, Facebook, Instagram, and YouTube. The inventory search page adds ItemList schema with individual vehicle listings (position, name, VIN, image). However, the department schema types are inconsistent: Sales is typed as AutoBodyShop rather than AutoDealer, and Parts is also AutoBodyShop rather than AutoPartsStore. The Ricart Ford subdomain (ricartford.com) uses the more appropriate AutoDealer type with richer sameAs including DealerRater, Cars.com, CarGurus, and LinkedIn -- signals the main site lacks.

External Signals and Fragmentation

The brand operates a fragmented multi-domain architecture: ricart.com (umbrella), ricartford.com, ricarthyundai.com, ricartused.com, ricartparts.com, ricarttobusiness.com, farrowhd.com, rabidcustoms.com, ricartexpressnewark.com, ricartexpresssunbury.com. Each subdomain has its own schema, sitemap, and content -- but the main site's external links point to these domains, creating a citation fragmentation problem for AI engines trying to consolidate the brand. The "Top Place to Work" page references a 2016 Columbus CEO Magazine article, suggesting stale content. The "About" page says "Celebrating 72 years" (founded 1953 = 72 years in 2025), which is current. The community-giving page lists dozens of charity partners but no recent dates or news items.

Findings

1. Cold LLM knowledge references 'no-haggle' pricing not present on site Medium

   LLM prior knowledge describes Ricart as having 'no-haggle' pricing, but the site never uses that term; instead it promotes a 'Best Value Guarantee' and '72 Hour Exchange Policy'.

   What to change: Add 'no-haggle' pricing language to the homepage and buying-power page to align with existing brand perception.

2. Cold LLM knowledge incorrectly lists Mazda as a brand sold Medium

   LLM prior knowledge states Ricart sells Mazda, but the site lists only Buick, Chevrolet, Ford, Genesis, GMC, Hyundai, Kia, Mitsubishi, and Nissan.

   What to change: Ensure the site clearly lists all brands sold and consider updating external sources to correct the Mazda reference.

3. llms.txt file returns 404 Medium

   The standard llms.txt file is missing, returning a branded 404 page. This file helps AI crawlers discover key content and context.

   What to change: Create an llms.txt file at the root listing important pages and a brief brand summary.

4. Department schema types are inconsistent and incorrect Medium

   Sales and Parts departments are typed as 'AutoBodyShop' instead of 'AutoDealer' and 'AutoPartsStore', respectively, which may confuse AI parsers.

   What to change: Update the JSON-LD to use 'AutoDealer' for Sales and 'AutoPartsStore' for Parts departments.

5. Fragmented multi-domain architecture dilutes brand authority High

   The brand operates multiple subdomains (ricartford.com, ricarthyundai.com, etc.) with separate schema and content, creating citation fragmentation for AI engines trying to consolidate brand information.

   What to change: Consolidate brand content under a single domain or use cross-domain schema references (sameAs) to signal brand unity.

6. Top Place to Work page references outdated 2016 article Low

   The 'Top Place to Work' page cites a 2016 Columbus CEO Magazine article, indicating stale content that may reduce credibility.

   What to change: Update the page with more recent awards or remove the date reference.

7. Community-giving page lacks recent dates or news items Low

   The community-giving page lists many charity partners but no recent dates or news, making it appear static.

   What to change: Add recent news or a blog section to show ongoing community involvement.

8. Robots.txt lacks AI-bot-specific directives Low

   The robots.txt uses a single User-Agent: * rule with no explicit allow/disallow for AI crawlers, leaving access ambiguous.

   What to change: Add explicit directives for AI crawlers (e.g., GPTBot, ClaudeBot) to ensure intended access.

9. Main site lacks richer sameAs links present on subdomain Medium

   The main site's sameAs includes only X, Facebook, Instagram, and YouTube, while the Ford subdomain includes DealerRater, Cars.com, CarGurus, and LinkedIn, which are stronger signals for AI.

   What to change: Add DealerRater, Cars.com, CarGurus, and LinkedIn to the main site's sameAs array.

What's working

• All major AI crawlers receive full HTML content — Every tested AI crawler (GPTBot, ClaudeBot, etc.) gets a 200 status with full HTML, no blocking or JS shells.
• AutomotiveBusiness JSON-LD with complete details on all pages — Every page carries AutomotiveBusiness schema with address, geo, phone, hours, departments, and sameAs links.
• Inventory search page includes ItemList schema with vehicle details — The searchnew.aspx page uses ItemList schema listing individual vehicles with position, name, VIN, and image.
• Dedicated EV center page with rich content — The /ricart-ev-center page provides 1396 words of EV-specific content, helping AI understand the brand's EV offerings.
• Service page with detailed content and schema — The /ricart-service page has 1302 words of service information, supporting AI understanding of service offerings.
• Buying Power page explains exchange and guarantee policies — The /ricart-buying-power page details the Best Value Guarantee and 72 Hour Exchange Policy, providing clear policy information for AI.
• About page correctly states 72 years of operation — The about page says 'Celebrating 72 years', which aligns with the 1953 founding date, showing current content.
• Ford subdomain uses AutoDealer type with richer sameAs — The ricartford.com subdomain uses the more appropriate AutoDealer schema type and includes additional sameAs links to DealerRater, Cars.com, etc.