mosescars.com — AI Site Grade

Analysis

I have enough data to write a sharp audit. Let me compile.

The Cold-Knowledge Gap

The most striking finding is that a frontier LLM, queried cold, describes Moses Auto Group as a used car dealership focused on "pre-owned vehicles" and "in-house financing for buyers with less-than-perfect credit" — while the site itself is a sprawling 16-location, 12-brand new-car auto group that has been operating since 1947 (or 1943, depending on the page). The model has no awareness that Moses sells new BMW, Cadillac, Chevrolet, Ford, GMC, Honda, Lexus, Lincoln, Nissan, Toyota, Volkswagen, or Mitsubishi vehicles. This is a severe positioning failure: the brand's actual identity as a multi-franchise new-car powerhouse is invisible to AI engines.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Applebot-Extended, Bytespider, anthropic-ai, ChatGPT-User — receive a 200 with full HTML content on the homepage. No UA-based blocking, no Cloudflare challenge, no JS shell. The robots.txt uses a single User-Agent: * rule with no AI-specific directives; GPTBot and friends are implicitly allowed everywhere except a handful of internal ASPX endpoints. The site runs on nginx behind Varnish cache (DealerOn platform), with a 4-hour TTL and generous stale-while-revalidate window. /llms.txt returns a 404 — the site has no AI-friendly content map.

Schema Posture

Every page carries the same AutomotiveBusiness JSON-LD block with address, geo, hours, and department sub-entities (AutoDealer, AutoRepair, AutoBodyShop). The schema is technically valid but generic: the @type is AutomotiveBusiness rather than AutoDealer at the top level, the description field contains a placeholder string ("Group dealer in St. Albans, WV. We specialize in new Group and used vehicles"), and the sameAs array includes a dead Twitter link (https://www.twitter.com/#). No Vehicle or Car schema is present on inventory pages — the 2,400+ vehicle detail URLs in the sitemap carry no structured data about make, model, price, or mileage.

Stale Content Signals

The sitemap includes a /covid-19.html page with an update timestamp of September 1, 2020 — pandemic-era operating hours that are now five years stale. The page is indexed and discoverable. The value-your-trade.html URLs in the sitemap contain malformed numeric suffixes (e.g., value-your-trade.html12772, value-your-trade.html6310), suggesting broken internal linking or template artifacts. The copyright footer reads "Copyright 2026 by DealerOn" — a forward-dated year that undermines trust signals.

External Signals

Web searches for "Moses Auto Group" and "Moses Cars St. Albans" return zero results on DuckDuckGo — no reviews, no news, no Reddit threads, no press mentions surfaced. The DNS TXT records show a google-site-verification token and a Zendesk/Outlook SPF record, confirming the site uses Zendesk for support. The brand has no detectable off-domain footprint that AI engines could cite, which compounds the cold-knowledge gap: the LLM's sparse, inaccurate prior is the only signal available.

Findings

1. Cold-knowledge describes Moses as a used-car dealership, missing 12 new-car brands High

   A frontier LLM queried cold describes Moses Auto Group as a used car dealership focused on pre-owned vehicles and in-house financing, while the site actually represents a 16-location, 12-brand new-car auto group. The brand's actual identity as a multi-franchise new-car powerhouse is invisible to AI engines.

   What to change: Publish an /llms.txt file and a structured brand fact sheet that explicitly lists all new-car franchises, locations, and history. Ensure the homepage and about pages clearly state the full scope of operations.

2. No /llms.txt file published High

   The site returns a 404 for /llms.txt, meaning there is no AI-friendly content map to guide language models to accurate brand information.

   What to change: Create an /llms.txt file that provides a concise, factual overview of the business, including all new-car brands, locations, and key differentiators.

3. JSON-LD uses generic AutomotiveBusiness type instead of AutoDealer Medium

   Every page carries the same AutomotiveBusiness JSON-LD block with a placeholder description and a dead Twitter link. The top-level type should be AutoDealer to better represent the business.

   What to change: Change the @type to AutoDealer, update the description to accurately reflect the full new-car inventory, and fix or remove the dead Twitter link.

4. No Vehicle or Car schema on inventory pages High

   The 2,400+ vehicle detail URLs in the sitemap carry no structured data about make, model, price, or mileage. This prevents AI engines from extracting specific inventory information.

   What to change: Add Vehicle or Car schema markup to all inventory detail pages, including make, model, year, price, mileage, and VIN.

5. Stale COVID-19 page with 2020 operating hours Medium

   The sitemap includes a /covid-19.html page with an update timestamp of September 1, 2020, containing pandemic-era operating hours that are now five years stale.

   What to change: Remove or update the COVID-19 page with current information, or redirect it to a relevant page.

6. Malformed numeric suffixes on value-your-trade URLs Low

   The sitemap contains URLs like value-your-trade.html12772 and value-your-trade.html6310, suggesting broken internal linking or template artifacts.

   What to change: Clean up the sitemap to remove malformed URLs and fix the underlying template issue.

7. Copyright footer reads 2026, undermining trust Low

   The copyright footer displays 'Copyright 2026 by DealerOn', a forward-dated year that may reduce trust signals for AI engines and users.

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

8. No detectable off-domain footprint in web search results High

   Web searches for 'Moses Auto Group' and 'Moses Cars St. Albans' return zero results on DuckDuckGo, meaning no reviews, news, or press mentions are surfaced. This compounds the cold-knowledge gap.

   What to change: Build an off-domain presence through Google Business Profile, review platforms, local citations, and press releases to create a verifiable external footprint.

9. Dead Twitter link in schema sameAs array Low

   The JSON-LD schema includes a sameAs link to https://www.twitter.com/#, which is a dead link and provides no value.

   What to change: Remove the dead Twitter link or replace it with the actual social media profile URL.

What's working

• All major AI crawlers receive full HTML content — All tested AI crawlers (GPTBot, ClaudeBot, PerplexityBot, etc.) receive a 200 status with full HTML content on the homepage, with no UA-based blocking or JavaScript shells.
• robots.txt implicitly allows all AI bots — The robots.txt uses a single User-Agent: * rule with no AI-specific directives, meaning GPTBot and other AI crawlers are allowed everywhere except a few internal ASPX endpoints.
• Valid AutomotiveBusiness JSON-LD on every page — Every page carries a technically valid AutomotiveBusiness JSON-LD block with address, geo, hours, and department sub-entities, providing basic location and contact information.
• Sitemap includes 2,400+ vehicle detail URLs — The sitemap contains over 2,400 vehicle detail URLs, providing a comprehensive inventory map for crawlers.
• Fast caching with Varnish and stale-while-revalidate — The site uses nginx behind Varnish cache with a 4-hour TTL and generous stale-while-revalidate window, ensuring fast page loads for crawlers and users.