bergstromauto.com — AI Site Grade

Analysis

I have a comprehensive picture now. Let me write the audit.

Bergstromauto.com: AI-Visibility Audit

The site's robots.txt treats GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and every other AI crawler identically to a browser — no AI-specific rules exist, and all bots receive full 200-status HTML with ~415KB of content — yet the site has no llms.txt (404s), no AI-friendly content map, and its JSON-LD schema describes the business as a "Chevrolet dealer in Neenah, WI" while the site itself claims to be "Wisconsin's #1 Automotive Marketplace" with 34 brands and 50+ locations.

Crawler Access

All 11 tested AI user-agents (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, Bytespider, Applebot-Extended, anthropic-ai, Perplexity-User) return 200 status with ~415KB of content — identical to browser baseline. No UA-based blocking, no Cloudflare challenge, no JS shell. The site runs on nginx behind Varnish cache, hosted on DealerOn platform (DNS: dns1.dealeron.com, dns2.dealeron.com). Cache headers (max-age=14400, stale-while-revalidate=1209600) are generous. The robots.txt has a single User-Agent: * rule with Crawl-delay: 10 and no AI-bot directives whatsoever — a missed opportunity to prioritize or segment crawler traffic. /llms.txt returns a 404 (the DealerOn 404 page, 61KB of HTML).

Schema Posture

Every page carries the same AutomotiveBusiness JSON-LD block with a description that reads: *"Bergstrom Automotive is a Chevrolet dealer in Neenah, WI. We specialize in new Chevrolet, used cars, service, and financing."* This is stale and reductive — the homepage and about page describe the company as operating 50+ locations across Wisconsin, representing 34 brands, with $2 billion in sales. The schema describes a single Chevrolet dealership. The FAQ page does include a FAQPage schema with 7+ Question/Answer pairs, which is a strong answer-format signal. No Product schema on vehicle inventory pages was verified, but the site's search result pages (.aspx URLs) are likely dynamically generated and may lack structured data for individual vehicles.

Cold-Knowledge Gap

The LLM prior knows Bergstrom as a "large, family-owned auto dealership group based in Wisconsin, operating over 30 dealerships" with "Ford, Chevrolet, Toyota, Honda, Subaru, and luxury marques." It mentions the "Bergstrom Difference," Milwaukee Brewers sponsorship, and community work. The site itself claims 50+ locations, 34 brands, 4,535 new cars, 1,467 pre-owned cars, 20,000+ 5-star Google reviews, and a Forbes Top 500 Midsize Employer ranking. The gap: the LLM prior undercounts locations (30 vs 50+), omits the upfront-pricing/no-negotiation model entirely, and does not know about the 7-day return policy, Price Protection Guarantee, or the fact that sales staff are paid on volume/guest satisfaction rather than commission. These are the site's strongest differentiators and they are invisible to cold AI knowledge.

Content & Signals

The homepage delivers 1,549 words of visible text with clear H1/H2 structure, upfront pricing messaging, and embedded customer reviews. The FAQ page contains 2,072 words of Q&A content with FAQPage schema. The news page shows 18 press releases dating back to 2023, including recent accolades (Forbes Top 500 Midsize Employer, Great Place to Work certification, Automotive News top dealership group). The "Why Buy" page explicitly states the three guarantees: Upfront Pricing, 7-Day Lowest Price Protection, and 7-Day Money-Back Guarantee. The about page lists executive leadership and community partnerships (Make-A-Wish, United Way, Drive for a Cure). The careers page notes 2,500+ team members and ranks among top 50 U.S. auto retailers.

External Signals

DNS records show Google Workspace (MX: aspmx.l.google.com), Ultipro (SPF include), Zoom, Cisco, and Duo SSO verification — indicating a mature enterprise tech stack. The site is built on DealerOn, a common automotive dealer CMS platform. No major external controversy surfaced in search. The news page links to a BizTimes profile, a Car Dealership Guy podcast, and an Automotive News "Daily Drive" podcast featuring the executive chairman — all strong third-party credibility signals that AI engines can cite.

Surprising Findings

The schema description mismatch is the most consequential finding: every page on the site tells AI crawlers that Bergstrom is a single Chevrolet dealer, while the actual content describes a multi-brand, multi-location group. This directly undermines how AI models classify and rank the business. The missing llms.txt (404) is a lost opportunity for a site with this much structured inventory data. The sitemap contains 6,227 URLs — mostly brand-specific landing pages (/new-chevrolet, /new-bmw, etc.) and search filter URLs — but the site's .aspx vehicle detail pages may not all be indexed efficiently. The copyright footer reads "2026" — either a forward-dated placeholder or a DealerOn default that has not been updated.

Findings

1. Schema describes Bergstrom as a single Chevrolet dealer, not a multi-brand group High

   Every page carries AutomotiveBusiness JSON-LD stating 'Bergstrom Automotive is a Chevrolet dealer in Neenah, WI' while the site claims 50+ locations, 34 brands, and $2B in sales. This mismatch misleads AI crawlers about the business scope.

   What to change: Update the JSON-LD @type to AutoDealer or AutomotiveBusiness with a description that reflects the multi-brand, multi-location group, and add Place schema for each location.

2. No llms.txt file published Medium

   The site returns a 404 for /llms.txt, missing an opportunity to provide AI crawlers with a curated content map and structured data about inventory, locations, and policies.

   What to change: Create an llms.txt file listing key pages (upfront pricing, FAQ, news, careers, inventory) and include a link to the sitemap.

3. Key differentiators invisible to cold AI knowledge High

   LLM prior knowledge omits upfront pricing, no-negotiation model, 7-day return policy, and Price Protection Guarantee. These are the site's strongest selling points and are not surfaced in AI training data.

   What to change: Publish an llms.txt and ensure schema markup on key pages explicitly states these policies (e.g., using WarrantyScope or ServicePromise schema).

4. Robots.txt lacks AI-specific directives Low

   The robots.txt has a single User-Agent: * rule with Crawl-delay: 10 and no AI-bot segmentation. This misses the chance to prioritize or guide AI crawlers to important pages.

   What to change: Add specific directives for GPTBot, ClaudeBot, and other AI crawlers, allowing full access but with a lower crawl-delay if needed.

5. Copyright footer displays year 2026 Low

   The footer reads '2026 Bergstrom Automotive' which is either a forward-dated placeholder or a DealerOn default that has not been updated, potentially signaling outdated content to crawlers.

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

6. Vehicle inventory pages likely lack Product schema Medium

   The site's search result pages use .aspx URLs and are dynamically generated; individual vehicle detail pages may not include Product or Vehicle structured data, reducing AI understanding of inventory.

   What to change: Add Product and Vehicle schema markup to all vehicle detail pages, including make, model, year, price, and availability.

What's working

• All major AI crawlers receive full HTML content — All 11 tested AI user-agents return 200 status with ~415KB of content, identical to browser baseline. No blocking or JS shells.
• FAQ page includes FAQPage schema with Q&A pairs — The FAQ page contains 2,072 words of Q&A content with FAQPage JSON-LD markup, providing strong answer-format signals for AI.
• Rich content with third-party credibility signals — The site features press releases, podcast appearances, and awards (Forbes Top 500 Midsize Employer, Great Place to Work) that AI engines can cite as authoritative sources.
• Cache headers allow efficient crawling — Cache-Control: max-age=14400 and stale-while-revalidate=1209600 reduce server load and improve crawl efficiency.