islandautogroup.com — AI Site Grade

Analysis

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

The key findings are: 1. The site is behind Akamai CDN which blocks ALL browser UAs and ALL AI bots EXCEPT anthropic-ai (which gets through to the nginx backend) 2. The robots.txt explicitly blocks GPTBot, OAI-SearchBot, ChatGPT-User, Claude-User, Claude-SearchBot, PerplexityBot, etc. from accessing JS/CSS/JSON/API paths 3. The site has an llms.txt (708KB) — a very large one — which is unusual and interesting 4. The site runs on dealer.com (DDC) platform 5. Cold LLM knowledge mentions "Island Ford" and "Island Hyundai" as separate entities, but the actual site sells Kia, Dodge, Hyundai, Honda, Mazda, Chrysler, Chevrolet, Jeep, Subaru, Toyota, Nissan, Volkswagen, Ram — no Ford mentioned 6. No JSON-LD schema visible in the fetched content 7. About-us page returns 404 8. No external search results found for the brand

Let me write the audit.

The Audit

Island Auto Group's site is fully invisible to every major AI crawler except one — and that one exception (anthropic-ai) is the only reason the site has any AI visibility at all. The Akamai CDN edge returns 403 to GPTBot, Google-Extended, OAI-SearchBot, PerplexityBot, ClaudeBot, and even standard browser User-Agents. Only anthropic-ai reaches the nginx backend and receives a 200 with full HTML content.

Crawler Access

The robots.txt is itself blocked behind the Akamai 403 wall for all standard UAs, but when fetched as anthropic-ai it reveals explicit restrictions: GPTBot, OAI-SearchBot, ChatGPT-User, Claude-User, Claude-SearchBot, PerplexityBot, CCBot, and Bytespider are all disallowed from /*.js, /*.css, /*.json, /api/, /apis/, and /pixall/. No Disallow: / is present for these bots — the core pages are technically crawlable — but the Akamai edge layer preemptively blocks them all at 403 regardless. The site runs on the dealer.com (DDC) platform behind AkamaiGHost, with a secondary nginx origin server.

Cold-Knowledge Gap

A frontier LLM queried cold describes Island Auto Group as selling Ford, Hyundai, and Kia and mentions separate "Island Ford" and "Island Hyundai" locations. The actual site sells Kia, Dodge, Hyundai, Honda, Mazda, Chrysler, Chevrolet, Jeep, Subaru, Toyota, Nissan, Volkswagen, and Ram — no Ford brand anywhere. The model's prior is fragmented and outdated, referencing dealership sub-brands that do not appear on the current site. The site's own homepage title targets "New & Used Cars near Jersey City" and positions itself as "the number one car dealership in Staten Island," but the model's cold knowledge does not reflect this claim.

Schema and Content Posture

The homepage HTML contains no JSON-LD structured data of any kind — no AutoDealer, Organization, LocalBusiness, or Product schema. The about-us.htm page returns a 404 (rendered as a branded "Oops!" page with index, follow meta). The llms.txt file exists and is unusually large (708KB), listing dozens of inventory and service pages with descriptions — a strong signal for AI discoverability that is undercut by the Akamai wall blocking most AI crawlers from reaching those pages. The site uses ddc-site CSS classes and pictures.dealer.com for image hosting, confirming the dealer.com platform.

External Signals

No external search results surfaced for the brand across multiple queries — no reviews, no Reddit threads, no press mentions, no dealer listings on third-party sites. The brand has effectively zero off-domain footprint in current search indexes, which compounds the AI visibility problem: AI models have no external signals to triangulate against when the site itself is walled off.

Findings

1. Akamai CDN blocks all AI crawlers except anthropic-ai High

   The Akamai edge returns 403 for GPTBot, Google-Extended, OAI-SearchBot, PerplexityBot, ClaudeBot, and standard browser UAs. Only anthropic-ai reaches the nginx backend with a 200 response.

   What to change: Configure Akamai to allow major AI crawlers (GPTBot, Google-Extended, OAI-SearchBot, PerplexityBot, ClaudeBot) to access the site, or serve them static HTML versions.

2. Robots.txt blocks AI bots from JS, CSS, JSON, and API paths Medium

   The robots.txt explicitly disallows GPTBot, OAI-SearchBot, ChatGPT-User, Claude-User, Claude-SearchBot, PerplexityBot, CCBot, and Bytespider from accessing /*.js, /*.css, /*.json, /api/, /apis/, and /pixall/ paths.

   What to change: Remove or relax the Disallow rules for AI bots on JS/CSS/JSON/API paths, or ensure those resources are not needed for content extraction.

3. No JSON-LD structured data on homepage High

   The homepage HTML contains no JSON-LD structured data for AutoDealer, Organization, LocalBusiness, or Product schema, which limits AI understanding of the business.

   What to change: Add JSON-LD structured data for AutoDealer, LocalBusiness, and Inventory listings on relevant pages.

4. About-us page returns 404 Medium

   The /about-us.htm page returns a 404 error (branded 'Oops!' page), preventing AI crawlers from accessing company information.

   What to change: Restore the about-us page with accurate company information and ensure it returns a 200 status.

5. Zero external search results for the brand High

   Multiple web searches for 'Island Auto Group' and related terms returned no results, indicating no reviews, listings, or mentions in search indexes.

   What to change: Build external citations through Google Business Profile, dealer directories, and local SEO efforts to create off-domain signals.

6. Cold LLM knowledge mismatches actual inventory Medium

   A frontier LLM's cold knowledge describes Island Auto Group as selling Ford, Hyundai, and Kia, but the actual site sells Kia, Dodge, Hyundai, Honda, Mazda, Chrysler, Chevrolet, Jeep, Subaru, Toyota, Nissan, Volkswagen, and Ram — no Ford.

   What to change: Ensure the site clearly lists all brands and models with structured data to correct LLM knowledge.

7. Large llms.txt exists but most AI bots cannot access it Medium

   The llms.txt file is 708KB and lists many pages, but since most AI bots are blocked by Akamai, they cannot reach the referenced content.

   What to change: Allow major AI crawlers through Akamai to make the llms.txt content accessible.

8. Robots.txt blocked for standard UAs Medium

   The robots.txt returns 403 for standard browser User-Agents, preventing search engines from reading crawl directives.

   What to change: Configure Akamai to allow standard search engine bots to access robots.txt.

9. Sitemap.xml returns 403 for crawlers High

   The sitemap.xml returns 403 when fetched, preventing search engines and AI crawlers from discovering the site's URL structure.

   What to change: Allow access to sitemap.xml for all crawlers by adjusting Akamai rules.

10. No llms.txt at domain root Low

    The llms.txt file is only accessible at www subdomain; the root domain returns 403.

    What to change: Add an llms.txt at the root domain or redirect to the www version.

What's working

• Anthropic-ai crawler allowed through Akamai — The anthropic-ai bot is permitted through the Akamai edge and receives full HTML content, providing some AI visibility.
• Large llms.txt file with many pages listed — The llms.txt file is 708KB and lists dozens of inventory and service pages with descriptions, aiding AI discoverability for allowed bots.
• Core pages return 200 for anthropic-ai — Key pages like homepage, new/used inventory, service, promotions, and contact return 200 status when fetched as anthropic-ai, with full HTML content.
• Site runs on dealer.com (DDC) platform — The site uses the dealer.com platform, which is a known automotive CMS with standard inventory management features.