ankorstore.com — AI Site Grade

Analysis

Ankorstore's robots.txt issues a blanket Disallow: / to all crawlers, yet Google-Extended, PerplexityBot, and Perplexity-User all receive full 200 responses with real content — a contradiction that leaves AI visibility governed by ad-hoc WAF rules rather than declared policy.

Crawler Access

The robots.txt at ankorstore.com/robots.txt contains a single rule: User-agent: * Disallow: /. No AI bot is explicitly mentioned. Despite this blanket block, compare_bot_access reveals that Google-Extended, PerplexityBot, and Perplexity-User all receive 200 status with ~927KB of content — identical to a browser baseline. Meanwhile, GPTBot, ClaudeBot, OAI-SearchBot, ChatGPT-User, anthropic-ai, and Bytespider all get 403 Forbidden (134-byte response). The site runs on Google Cloud infrastructure (istio-envoy, Google Cloud DNS, Google-hosted sitemaps) with no Cloudflare layer. The WAF or ingress rules selectively block certain AI UAs while letting others through, creating a de facto access policy that contradicts the robots.txt. No llms.txt exists (404).

Cold-Knowledge Gap

A frontier LLM queried cold about Ankorstore recalls the company as a French B2B wholesale marketplace founded in 2019 by ex-Aircall and Algolia employees, with over EUR 250 million raised by 2022, and mentions layoffs and strategic struggles in 2023-2024. The actual site tells a different story: the about page claims EUR 365 million raised (not 250M), 30,000 partner brands, 300,000 registered retailers, 140 employees, and coverage across 30 European countries. The cold model knows nothing about the current scale (300K retailers, 30K brands) and cites an outdated funding figure. The model's knowledge of "reputational challenges and layoffs" is entirely absent from the site's own narrative, which presents only growth metrics and mission language.

Schema Posture

The homepage and category pages carry a single Organization schema with sameAs links to 8 social profiles and an AggregateRating (4.1/5 from 1,013 reviews). No WebSite, SearchAction, ItemList, or Product schema is present on any page examined. Category pages like /shop/home-kitchen-1246 list dozens of products with names, brands, and prices but mark them as plain H2 headings with zero structured data. The about page at my.ankorstore.com/about-us/en has no JSON-LD at all. The sitemap index reveals over 14,000 brand pages and 100+ product sub-sitemaps, yet none of these product pages were sampled to carry Product schema — a massive missed signal for AI answer engines.

External Signals

Web searches for Ankorstore reviews, Reddit discussions, and news about layoffs returned zero results from the DuckDuckGo index. The DNS TXT records show an anthropic-domain-verification token, indicating the brand has proactively registered with Anthropic for Claude crawler access — yet ClaudeBot receives a 403 Forbidden on the homepage. The site has multiple Google Search Console verification records and an Applebot verification token, suggesting awareness of crawler management, but the execution is fragmented: some bots are verified and blocked simultaneously.

Findings

1. Robots.txt blocks all crawlers but WAF selectively allows some AI bots High

   The robots.txt issues a blanket Disallow: / to all user agents, yet Google-Extended, PerplexityBot, and Perplexity-User receive full 200 responses with real content. GPTBot, ClaudeBot, and others get 403 Forbidden. This contradiction means AI visibility is governed by ad-hoc WAF rules rather than a declared policy.

   What to change: Update robots.txt to explicitly allow beneficial AI crawlers (e.g., Google-Extended, PerplexityBot) and align WAF rules with the declared policy. Remove the blanket Disallow: / or replace it with targeted rules.

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 list of important pages and context.

   What to change: Create an llms.txt file that lists key pages (e.g., about, category pages, popular brands) and provides a brief description of the marketplace.

3. Cold LLM knowledge omits current scale and uses outdated funding figure High

   A frontier LLM queried cold recalls Ankorstore as a French B2B wholesale marketplace with EUR 250 million raised, but the site claims EUR 365 million raised, 30,000 brands, 300,000 retailers, and 140 employees. The model knows nothing about the current scale and mentions layoffs absent from the site's narrative.

   What to change: Ensure key pages (about, homepage) contain up-to-date facts in clear prose and structured data. Consider publishing a press kit or fact sheet that AI crawlers can easily index.

4. Product pages lack Product structured data High

   Category pages list dozens of products with names, brands, and prices but mark them as plain H2 headings with zero structured data. The sitemap index reveals over 14,000 brand pages and 100+ product sub-sitemaps, yet none of the sampled pages carry Product schema.

   What to change: Add Product JSON-LD schema to all product listing and detail pages, including name, brand, price, availability, and aggregate rating where applicable.

5. Homepage and category pages lack WebSite and SearchAction schema Medium

   The homepage and category pages carry only Organization schema. No WebSite, SearchAction, ItemList, or BreadcrumbList schema is present, limiting visibility in AI answer engines.

   What to change: Add WebSite schema with SearchAction on the homepage and category pages. Add ItemList schema on category pages to enumerate products.

6. About page has no structured data at all Medium

   The about page at my.ankorstore.com/about-us/en contains no JSON-LD, missing an opportunity to provide AI crawlers with verified organization details.

   What to change: Add Organization schema to the about page with all relevant fields: name, description, founding date, number of employees, funding, and social profiles.

7. ClaudeBot blocked despite anthropic-domain-verification DNS record High

   The DNS TXT records include an anthropic-domain-verification token, indicating proactive registration with Anthropic, yet ClaudeBot receives a 403 Forbidden on the homepage.

   What to change: Allow ClaudeBot access by updating WAF rules and robots.txt to permit the anthropic-ai user agent.

8. GPTBot receives 403 Forbidden High

   GPTBot is blocked with a 403 response, preventing OpenAI's crawler from indexing the site.

   What to change: Allow GPTBot access by updating WAF rules and robots.txt to permit the GPTBot user agent.

9. No external reviews or Reddit discussions indexed in search Medium

   Web searches for Ankorstore reviews, Reddit discussions, and news about layoffs returned zero results from the DuckDuckGo index, indicating low external signal presence.

   What to change: Encourage customer reviews on third-party platforms and engage in relevant online communities to build external signals.

10. Shop page has very little content Low

    The /shop page returns only 155 words, providing minimal context for AI crawlers to understand the marketplace.

    What to change: Expand the /shop page with descriptive text about the marketplace, categories, and how to use the platform.

What's working

• Homepage carries Organization schema with aggregate rating — The homepage includes Organization JSON-LD with sameAs links to 8 social profiles and an AggregateRating of 4.1/5 from 1,013 reviews, providing basic structured data for AI crawlers.
• Google-Extended crawler receives full content — Google-Extended is allowed through the WAF and receives a 200 response with full page content, enabling Google's AI to index the site.
• PerplexityBot and Perplexity-User receive full content — Both PerplexityBot and Perplexity-User are allowed through the WAF and receive 200 responses with full page content, enabling Perplexity's AI to index the site.
• Sitemap index is well-organized with multiple sub-sitemaps — The sitemap index contains 80 URLs and references sub-sitemaps for brands and products, indicating a structured approach to site navigation.
• Brand pages contain substantial product descriptions — A sampled brand page (La Pousse Verte) contains 1769 words with product descriptions, providing rich content for AI crawlers.
• DNS records include verification tokens for multiple crawlers — The DNS TXT records contain anthropic-domain-verification, google-site-verification, and apple-domain-verification tokens, showing proactive crawler management efforts.