dutchmillerauto.com — AI Site Grade

Analysis

I have enough data now. Let me compile the audit.

The most striking finding: the bare domain (dutchmillerauto.com) has a broken SSL certificate — it redirects to www but the cert is invalid for the bare domain, meaning any AI crawler hitting the canonical domain directly gets a TLS error. Meanwhile, the www subdomain works fine and is fully open to all AI bots.

Let me write the audit.

Dutch Miller Auto Group — AI-Visibility Audit

The bare domain dutchmillerauto.com presents a broken SSL certificate (hostname mismatch), causing every AI crawler that lands on the canonical domain to receive a TLS connection error — yet the www subdomain serves all bots identically with full content, creating a split-personality crawl surface that likely fragments AI ingestion.

Crawler Access

All 11 tested AI bot user-agents (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, Bytespider, Applebot-Extended, anthropic-ai, Perplexity-User) receive a 200 OK with ~370KB of content from https://www.dutchmillerauto.com — identical to browser baseline. No UA-based blocking exists. However, the robots.txt contains zero AI-bot-specific directives; a single catch-all User-Agent: * rule with a Crawl-delay: 10 and standard ASP.NET disallows. No llms.txt exists (returns a 404 with a full HTML page shell). The site runs on nginx behind Varnish CDN with ASP.NET (DealerOn platform), and pages carry a 4-hour cache with 14-day stale-while-revalidate — meaning AI crawlers may serve stale inventory data for up to two weeks.

Cold-Knowledge Gap

The LLM knows Dutch Miller Auto Group as a family-owned West Virginia dealership network founded in the 1960s/1970s, selling Chevrolet, Buick, GMC, Ford, CDJR across multiple locations, with community involvement and mixed reviews. The actual site reveals a far larger operation: 16+ branded locations across WV, KY, NC, TN, and VA, employing ~300 people, selling Hyundai, Kia, Subaru, Nissan, Honda, and Lincoln alongside the expected brands. The LLM has no awareness of the Charlotte Kia location, the Ashland (KY) stores, or the Nissan/Bristol/Wytheville footprint. The cold knowledge also omits the group's tagline ("If You Don't Buy Dutch You Will Pay Way Too Much") and the founder's Marine Corps flag-raising story — both distinctive narrative assets the site does carry.

Schema Posture

Every page carries a single AutomotiveBusiness JSON-LD block with nested AutoDealer, AutoRepair, and AutoPartsStore departments. The schema is technically valid but contains broken sameAs links: https://www.twitter.com/DMKIACLT (a Kia-specific account, not the group) and https://www.facebook.com/dutchmiller.ofcharlotte (Charlotte-only, not the group). No AggregateRating schema exists despite the "largest volume dealer" claim. No FAQPage, Product, or Vehicle schema on inventory pages. The homepage title is a keyword-stuffed 150+ character string ("WV GMC dealer in Huntington WV - New and Used GMC dealership Ashland ky Hurricane wv ironton oh charleston wv huntington wv") that reads as spam to AI summarizers.

Content & Structural Issues

The about-us.html page returns a 404 — the actual about page is aboutus.aspx, creating a broken navigation path for any AI crawler following the natural /about-us convention. The community.html page contains only a single sentence of substantive content ("Here at Dutch Miller, we take great joy in serving our community") with no actual events listed. The career.html page is entirely empty of job listings — just the word "Careers" as an H1 with no content. No FAQ, comparison tables, or answer-format signals exist anywhere on the site. The sitemap contains 2,392 URLs, overwhelmingly individual vehicle detail pages — but no blog, no news, no educational content that an AI engine would cite as an authoritative source.

Findings

1. Bare domain SSL certificate hostname mismatch blocks AI crawlers High

   The bare domain dutchmillerauto.com presents a broken SSL certificate (hostname mismatch), causing every AI crawler that lands on the canonical domain to receive a TLS connection error. The www subdomain works correctly, creating a fragmented crawl surface.

   What to change: Fix the SSL certificate for the bare domain to match the hostname, or set up a proper redirect from the bare domain to the www subdomain with a valid certificate.

2. Robots.txt lacks AI-bot-specific directives Medium

   The robots.txt file contains only a single catch-all User-Agent: * rule with a Crawl-delay: 10 and standard ASP.NET disallows. No AI-bot-specific directives exist, leaving AI crawlers without guidance on which parts of the site to crawl or avoid.

   What to change: Add explicit allow/disallow rules for AI bots (e.g., GPTBot, ClaudeBot) to control crawl behavior and prevent unnecessary load.

3. No llms.txt file published Medium

   The llms.txt file returns a 404 with a full HTML page shell, meaning AI crawlers cannot discover a curated list of pages for LLM training or inference.

   What to change: Publish an llms.txt file listing key pages (about, inventory, service) to guide AI crawlers to the most important content.

4. Long cache TTL may serve stale inventory data to AI crawlers Medium

   Pages carry a 4-hour cache with 14-day stale-while-revalidate, meaning AI crawlers may serve stale inventory data for up to two weeks.

   What to change: Reduce the stale-while-revalidate period for inventory pages to ensure AI crawlers receive fresh data.

5. LLM knowledge of dealership network is incomplete and outdated High

   The LLM knows Dutch Miller Auto Group as a small family-owned West Virginia dealership, but the actual site reveals 16+ locations across five states, selling multiple brands not in the LLM's knowledge. The LLM has no awareness of the Charlotte Kia location, Ashland stores, or the Nissan/Bristol/Wytheville footprint.

   What to change: Publish structured data (LocalBusiness, AutoDealer) for each location with accurate name, address, and brand affiliations to update LLM knowledge.

6. JSON-LD sameAs links point to location-specific pages, not the group Medium

   The AutomotiveBusiness schema contains sameAs links to a Kia-specific Twitter account and a Charlotte-only Facebook page, not the group's official social media. This misrepresents the business to AI engines.

   What to change: Update the sameAs links to point to the group's official social media profiles (e.g., Dutch Miller Auto Group Facebook page).

7. No AggregateRating schema despite volume claims Medium

   The site claims to be the 'largest volume dealer' but does not include AggregateRating schema to back up reputation signals.

   What to change: Add AggregateRating schema with review count and average rating from Google or other review platforms.

8. Homepage title is keyword-stuffed and spammy Medium

   The homepage title is over 150 characters and reads as spam to AI summarizers: 'WV GMC dealer in Huntington WV - New and Used GMC dealership Ashland ky Hurricane wv ironton oh charleston wv huntington wv'.

   What to change: Rewrite the title to a concise, descriptive phrase under 60 characters, e.g., 'Dutch Miller Auto Group | GMC Dealer in Huntington, WV'.

9. About-us.html returns 404, breaking natural navigation High

   The page at /about-us.html returns a 404, while the actual about page is at /aboutus.aspx. AI crawlers following the natural /about-us convention hit a dead end.

   What to change: Redirect /about-us.html to /aboutus.aspx or create a proper page at /about-us.html.

10. Careers page contains no job listings Low

    The /career.html page has only an H1 'Careers' with no job listings or content, providing no value to AI crawlers or users.

    What to change: Add job listings or link to an external careers portal.

11. Community page lacks substantive content Low

    The /community.html page contains only a single sentence about serving the community with no actual events or details, offering little for AI to cite.

    What to change: Expand the community page with specific events, sponsorships, or volunteer activities.

12. No FAQ, comparison tables, or educational content for AI citation Medium

    The site lacks FAQ pages, comparison tables, or educational content that AI engines would cite as authoritative sources, reducing the likelihood of being referenced in AI responses.

    What to change: Create FAQ pages about common car-buying questions, service tips, and comparison guides to generate authoritative content.

What's working

• All 11 tested AI bots receive full content on www subdomain — All 11 tested AI bot user-agents receive a 200 OK with full content from https://www.dutchmillerauto.com, identical to browser baseline. No UA-based blocking exists.
• JSON-LD AutomotiveBusiness schema present on all pages — Every page carries a valid AutomotiveBusiness JSON-LD block with nested AutoDealer, AutoRepair, and AutoPartsStore departments, providing structured business information to AI engines.
• Sitemap contains 2,392 URLs with detailed vehicle inventory — The sitemap includes a large number of individual vehicle detail pages, providing rich inventory data for AI crawlers to index.
• Site runs on nginx behind Varnish CDN for performance — The site uses nginx behind Varnish CDN, ensuring fast page loads for both users and AI crawlers.
• Site carries distinctive brand narrative (founder's Marine Corps story, tagline) — The site includes the founder's Marine Corps flag-raising story and the tagline 'If You Don't Buy Dutch You Will Pay Way Too Much', providing unique content for AI to reference.
• Service page provides detailed service and parts information — The /service.html page contains 524 words of substantive content about service and parts, offering value to AI crawlers.
• Site has historical snapshots in Wayback Machine — The site has a Wayback Machine snapshot from 2026, indicating long-term web presence and archival stability.