plasticbank.com — AI Site Grade

Analysis

Plastic Bank's llms.txt is a 175KB firehose — and that is the most revealing thing about its AI posture

The site publishes a /llms.txt file weighing 175,006 bytes that dumps every page on the domain — including password-protected pages, checkout URLs, and author bios — into a single flat list with no hierarchy, no summaries, and no curation. This is the opposite of the llms.txt convention (which calls for a curated, concise AI-friendly content map). The file is so large and unfiltered that it undermines its own purpose: an AI crawler consuming it gets noise, not signal.

Crawler Access

Every major AI bot — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended, anthropic-ai, ChatGPT-User, Perplexity-User — receives a 200 status with full HTML content identical to browser delivery (429 KB). No UA-based blocking, no Cloudflare challenge, no JS shell. The robots.txt contains zero AI-specific directives — only generic WooCommerce admin exclusions. The site runs on Cloudflare with HSTS and standard security headers but does not discriminate between bots. This is technically permissive but strategically passive: all bots get the same full-page payload including login prompts, cart modals, and country-select dropdowns that add no semantic value.

Cold-Knowledge Gap

The LLM prior knows Plastic Bank as a Haiti-founded social enterprise with ~35,000 collectors and 100M+ kg collected, citing SC Johnson, Henkel, and Marks & Spencer as partners. The actual site tells a materially different story: 64,747 collectors, 194M+ kg collected (~9.7 billion bottles), operations in 5 countries (Philippines, Indonesia, Brazil, Egypt, Thailand) — Haiti is no longer listed. The model's knowledge is 2-3 years stale on every key metric. The model also recalls "plastic neutrality" criticism from 2022, but the site has fully rebranded around "Plastic Credits," "Impact Programs," and "Social Plastic feedstock" — the term "plastic neutrality" appears nowhere on the homepage or key product pages. The site has moved on; the model has not.

Schema Posture

Every page carries the same Organization + WebSite + WebPage + Article schema graph with identical @id references pointing to https://plasticbank.com/#organization. The schema is technically valid but generic: no Product schema for Plastic Credits or Social Plastic feedstock, no FAQPage schema on pages with FAQ sections, no BreadcrumbList on most pages (only on /about/ and /our-technology/). The pricing page has no Product or Offer schema despite listing tiered memberships with prices. The blog's comparison articles (Plastic Bank vs 4ocean, vs The Ocean Cleanup, vs WWF) have no Comparison schema markup.

Content & Answer-Format Signals

The site is unusually well-prepared for AI-driven discovery. The blog contains a dedicated "Comparison Guide" category with multi-thousand-word, table-rich, footnoted comparison articles against competitors (4ocean, The Ocean Cleanup, WWF, Waste Free Oceans). These are written by Gabriel Dalton, whose author bio reads "Generative Engine Optimization (GEO), Plastic Bank" — an explicit investment in AI search visibility. The homepage has FAQ accordions, the pricing page has comparison tables, and the technology page details blockchain, AI, and SaaS platform capabilities. The llms.txt includes every page but the blog's comparison articles are the strongest AI-answer-format content on the domain.

External Signals

The site lists partners including SC Johnson, Coca-Cola Philippines, Acer, CooperVision, Reckitt, Ocean Bottle, Charles & Keith, and Dr. Brown's. DNS records show integrations with Google Workspace, Mandrill, Stripe, Airtable, and 1Password. The domain has 5 Google site verification TXT records — suggesting multiple properties or historical migrations. No recent controversy surfaced in web searches; the 2022 "plastic neutrality" criticism the model recalls appears to have been addressed through rebranding to "Plastic Credits" and "Impact Programs." The site publishes annual sustainability reports (2021-2025) as downloadable PDFs, supporting transparency claims.

Findings

1. llms.txt is a 175KB unfiltered dump of all pages High

   The /llms.txt file contains 175,006 bytes listing every page on the domain, including password-protected pages, checkout URLs, and author bios, with no hierarchy or summaries. This violates the llms.txt convention and provides noise rather than signal to AI crawlers.

   What to change: Replace the current llms.txt with a curated, concise list of key pages (e.g., homepage, about, product pages, top blog posts) with brief summaries for each.

2. LLM knowledge is 2-3 years out of date on key metrics High

   The LLM prior recalls 35,000 collectors and 100M+ kg collected, while the site reports 64,747 collectors and 194M+ kg collected. Haiti is no longer listed as an operating country; the model still references it. The term 'plastic neutrality' has been replaced by 'Plastic Credits' and 'Impact Programs' on the site, but the model still associates criticism with that term.

   What to change: Publish an llms.txt with up-to-date key facts and consider adding a structured data fact sheet or knowledge panel to help LLMs refresh their knowledge.

3. No Product or Offer schema on pricing and product pages Medium

   The pricing page lists tiered memberships with prices but lacks Product and Offer schema. The Plastic Credits and Social Plastic feedstock pages also lack Product schema, reducing the chance of rich results in AI-driven search.

   What to change: Add Product and Offer schema to pricing and product pages, including price, currency, and availability.

4. Schema is generic Organization type across all pages Medium

   Every page uses the same Organization + WebSite + WebPage + Article schema graph with identical @id references. No FAQPage, BreadcrumbList, or Comparison schema is used where appropriate, limiting the richness of structured data.

   What to change: Add FAQPage schema to pages with FAQ sections, BreadcrumbList to all pages, and Comparison schema to comparison blog posts.

5. Comparison blog posts lack Comparison schema Medium

   The blog contains detailed comparison articles against competitors (4ocean, The Ocean Cleanup, WWF) but none use Comparison schema markup, missing an opportunity for rich results in AI-driven comparisons.

   What to change: Add Comparison schema to all comparison blog posts, referencing the compared entities and their attributes.

6. No AI-specific crawler optimization in robots.txt or headers Medium

   The robots.txt contains only generic WooCommerce exclusions and no AI-specific directives. All bots receive the same full HTML payload including non-semantic elements like login prompts and cart modals. The site does not leverage Cloudflare's bot management to prioritize AI crawlers.

   What to change: Add AI-specific directives in robots.txt to allow all AI bots, and consider using Cloudflare's bot management to serve a cleaner, semantic-only version to AI crawlers.

7. BreadcrumbList schema missing on most pages Low

   Only the /about/ and /our-technology/ pages have BreadcrumbList schema. Other pages lack it, reducing navigation clarity for AI crawlers.

   What to change: Add BreadcrumbList schema to all pages to improve navigation understanding for AI crawlers.

What's working

• Dedicated comparison guides for AI-driven discovery — The blog features a 'Comparison Guide' category with detailed, table-rich, footnoted articles comparing Plastic Bank to competitors, written by a GEO specialist. This content is highly suitable for AI answer generation.
• All major AI bots receive full HTML content — Every tested AI bot (GPTBot, ClaudeBot, etc.) receives a 200 status with full HTML content identical to browser delivery, with no blocking or challenges. This ensures maximum crawlability.
• Annual sustainability reports available as PDFs — The site publishes downloadable sustainability reports from 2021-2025, supporting transparency claims and providing authoritative content for AI crawlers.
• Generative Engine Optimization specialist on staff — Blog author Gabriel Dalton is listed as a GEO specialist, indicating explicit investment in AI search visibility and content optimization for AI-driven discovery.
• FAQ accordions on homepage provide answer-ready content — The homepage includes FAQ accordions that present common questions and answers in a format easily parsed by AI crawlers for direct answers.
• Technology page details blockchain, AI, and SaaS capabilities — The /our-technology/ page provides in-depth information about Plastic Bank's technology stack, including blockchain and AI, which can serve as authoritative content for AI crawlers.
• Partners page lists major global brands — The partners page includes well-known brands like SC Johnson, Coca-Cola Philippines, and Acer, providing strong external signals and credibility for AI crawlers.