kolarauto.com — AI Site Grade

Analysis

I have enough data to write the audit. Let me compile my findings.

Key observations: 1. Cloudflare blocks browser UAs but selectively allows ChatGPT-User, OAI-SearchBot, and Perplexity-User through 2. GPTBot, ClaudeBot, Google-Extended, PerplexityBot, Bytespider, Applebot-Extended all get 403 3. No AI-bot rules in robots.txt 4. No llms.txt (403) 5. Site is built on DealereProcess platform (dealereprocess.org CDN) 6. Heavy JS-rendered pages (~200KB each) 7. No JSON-LD schema detected 8. Cold LLM knowledge knows nothing about "Kolar Automotive" 9. No external search results found for the brand 10. Multi-brand dealership: Buick, Cadillac, Chevrolet, GMC, Hyundai, Toyota 11. Located in Hermantown, MN (near Duluth) 12. Phone: 218-733-0200

Let me write the audit.

Audit: kolarauto.com

The site's Cloudflare WAF blocks GPTBot, ClaudeBot, Google-Extended, PerplexityBot, Bytespider, and Applebot-Extended with 403 responses, while selectively allowing ChatGPT-User, OAI-SearchBot, and Perplexity-User through to full 200-byte pages. This creates a fragmented AI-crawler posture where OpenAI and Perplexity's user-facing agents can index the site but their core training crawlers cannot.

Crawler Access

The robots.txt at https://www.kolarauto.com/robots.txt contains no directives for any AI crawler — no GPTBot, ClaudeBot, Google-Extended, or PerplexityBot rules exist. The file only blocks generic scrapers (MJ12bot, BLEXBot, 008, DTS Agent) and internal functional paths. The llms.txt endpoint returns a Cloudflare 403 block. The sitemap at /resrc/xmlsitemap/xml-sitemaps/ is accessible to allowed bots and contains thousands of inventory URLs across six sub-sitemaps organized by location (Hermantown, Duluth, Superior, Minong, Ashland).

Content & Schema

The homepage is a heavy JavaScript-rendered page (~214KB) built on the DealereProcess platform (cdn.dealereprocess.org). The raw HTML contains a <title> tag and <meta name="description"> identifying Kolar Automotive Group as a multi-brand dealer (Buick, Cadillac, Chevrolet, GMC, Hyundai, Toyota) in Hermantown, MN near Duluth. No JSON-LD structured data of any kind was detected in the page source — no AutoDealer, LocalBusiness, Product, or Vehicle schema types are present. The visible text content is minimal in the raw HTML; most content is loaded client-side via JavaScript. Key pages include /search/new/tp/, /search/used/tp/, /about-us/, /service-department/, /parts-department/, /finance-department/, and /contact-us/.

Cold-Knowledge Gap

A frontier LLM queried cold about "Kolar Automotive" returned no verifiable information — the model could not confirm the brand's existence, products, or location. This is a complete knowledge vacuum. The site itself is a legitimate multi-franchise dealership with six OEM brands, a service department, parts, and financing, but AI models have zero prior awareness of it. The gap between the site's actual content (a full-service dealership group) and what AI engines know cold (nothing) is total.

External Signals

Web searches for kolarauto.com, "Kolar Automotive Group Hermantown", and "Kolar Automotive reviews" returned zero results across multiple queries. No Reddit threads, press mentions, review-site listings, or third-party citations were found. The brand has effectively no off-domain footprint that AI engines could surface through retrieval. The DNS records show the site uses Cloudflare for DNS and security, Microsoft 365 for mail (kolarauto-com.mail.protection.outlook.com), and KnowBe4 for security awareness training.

Findings

1. Cloudflare WAF blocks GPTBot, ClaudeBot, Google-Extended, and other core AI crawlers High

   The Cloudflare WAF returns 403 for GPTBot, ClaudeBot, Google-Extended, PerplexityBot, Bytespider, and Applebot-Extended, while allowing ChatGPT-User, OAI-SearchBot, and Perplexity-User. This fragmented access means training crawlers cannot index the site, limiting AI knowledge acquisition.

   What to change: Update Cloudflare WAF to allow GPTBot, ClaudeBot, Google-Extended, and other major AI training crawlers, or remove the bot-specific blocks to allow all legitimate crawlers.

2. Robots.txt contains no directives for AI crawlers High

   The robots.txt file at https://www.kolarauto.com/robots.txt has no rules for GPTBot, ClaudeBot, Google-Extended, or any other AI crawler. It only blocks generic scrapers and internal paths, leaving AI crawler access undefined.

   What to change: Add explicit allow or disallow rules for all major AI crawlers (GPTBot, ClaudeBot, Google-Extended, PerplexityBot, etc.) in robots.txt.

3. llms.txt endpoint returns 403 Medium

   The llms.txt file at https://kolarauto.com/llms.txt is blocked by Cloudflare with a 403 response. This prevents AI agents from discovering the site's intended AI-facing content and capabilities.

   What to change: Create and serve an llms.txt file at the root with a summary of the site and links to key pages, and ensure Cloudflare allows access to it.

4. No JSON-LD structured data on any page High

   No JSON-LD schema of any type (AutoDealer, LocalBusiness, Product, Vehicle, etc.) was detected in the page source of the homepage or any other crawled page. This prevents AI engines from extracting structured information about the dealership, its inventory, and services.

   What to change: Add JSON-LD structured data for AutoDealer, LocalBusiness, and Vehicle types on relevant pages, including name, address, phone, inventory, and operating hours.

5. Pages are heavy JavaScript-rendered shells Medium

   The homepage and other pages are approximately 214KB of HTML with minimal visible text content in the raw source; most content is loaded client-side via JavaScript. This can hinder AI crawlers that do not execute JavaScript, reducing the amount of indexable content.

   What to change: Implement server-side rendering or pre-rendering for key pages to ensure critical content is available in the initial HTML response.

6. No external search results or citations for the brand High

   Web searches for 'kolarauto.com', 'Kolar Automotive Group Hermantown', and related queries returned zero results. No review sites, press mentions, or third-party citations were found, giving AI retrieval systems no off-domain signals to surface.

   What to change: Build an off-domain presence through local business listings (Google Business Profile, Yelp, DealerRater), press releases, and social media profiles to generate indexed citations.

7. Frontier LLMs have no knowledge of Kolar Automotive High

   A cold query to a frontier LLM about 'Kolar Automotive' returned no verifiable information; the model could not confirm the brand's existence, products, or location. This indicates a complete knowledge vacuum despite the site being a legitimate multi-franchise dealership.

   What to change: Improve AI visibility by fixing crawler access, adding structured data, and building external citations so AI models can learn about the dealership.

8. Google-Extended crawler blocked by Cloudflare High

   Google-Extended, the crawler used for AI training by Google, receives a 403 from Cloudflare. This prevents Google from using the site's content to train its AI models, limiting the site's influence in AI-powered search features.

   What to change: Allow Google-Extended through the Cloudflare WAF to enable Google AI training on the site's content.

9. ClaudeBot crawler blocked by Cloudflare Medium

   ClaudeBot, Anthropic's AI training crawler, is blocked with a 403. This prevents Claude from learning about the dealership and its offerings.

   What to change: Allow ClaudeBot through the Cloudflare WAF to enable Anthropic AI training on the site's content.

10. PerplexityBot blocked while Perplexity-User allowed Medium

    PerplexityBot (the training crawler) is blocked with 403, but Perplexity-User (the user-facing agent) is allowed. This inconsistency means Perplexity's AI can answer user queries about the site but cannot train on it.

    What to change: Allow PerplexityBot through the Cloudflare WAF to enable Perplexity AI training on the site's content.

11. Applebot-Extended blocked by Cloudflare Low

    Applebot-Extended, Apple's AI training crawler, is blocked with a 403. This prevents Apple AI from indexing the site for training purposes.

    What to change: Allow Applebot-Extended through the Cloudflare WAF to enable Apple AI training on the site's content.

12. Bytespider blocked by Cloudflare Low

    Bytespider, ByteDance's AI training crawler, is blocked with a 403. This prevents TikTok/Douyin's parent company from training on the site's content.

    What to change: Allow Bytespider through the Cloudflare WAF to enable ByteDance AI training on the site's content.

What's working

• ChatGPT-User and OAI-SearchBot are allowed through Cloudflare — The Cloudflare WAF permits ChatGPT-User and OAI-SearchBot to access the site with full 200 responses, enabling OpenAI's user-facing agents to index and reference the site's content.
• Perplexity-User is allowed through Cloudflare — Perplexity-User receives a 200 response, allowing Perplexity's user-facing AI agent to access and surface the site's content in answers.
• Sitemap is accessible to allowed AI bots — The sitemap at /resrc/xmlsitemap/xml-sitemaps/ is accessible to ChatGPT-User and contains thousands of inventory URLs across six sub-sitemaps, helping allowed bots discover the site's content.
• Site clearly identifies as a multi-brand dealership with six OEM brands — The homepage title and meta description explicitly state Kolar Automotive Group sells Buick, Cadillac, Chevrolet, GMC, Hyundai, and Toyota in Hermantown, MN, providing clear brand identity for AI agents that can access the page.
• Key pages like inventory, about, service, parts, finance, and contact are accessible — Pages such as /search/new/tp/, /search/used/tp/, /about-us/, /service-department/, /parts-department/, /finance-department/, and /contact-us/ are all accessible to allowed bots, providing comprehensive information about the dealership.
• Cloudflare provides DNS and security protection — The site uses Cloudflare for DNS and security, which helps protect against DDoS and other attacks, ensuring site availability for allowed crawlers.
• Microsoft 365 email is configured — The site uses Microsoft 365 for email (kolarauto-com.mail.protection.outlook.com), indicating a professional email setup that can support business communications.