metrocu.org — AI Site Grade

Analysis

I have enough data to write a sharp audit. Let me compile.

The site's robots.txt has no AI-bot rules at all, yet every major AI crawler gets a full 200 with identical content — a rare open-door posture that is both an opportunity and a risk.

Crawler Access

All eleven tested AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, Bytespider, Applebot-Extended, anthropic-ai, Perplexity-User — receive a 200 with 296,739 bytes of identical HTML from the homepage, matching the browser baseline exactly. No UA-based blocking, no JS shell, no thin content. The robots.txt (453 bytes) contains a single User-agent: * rule disallowing only /cpresources/, /vendor/, /.env, and /cache/ — no AI-specific directives whatsoever. The site runs on Craft CMS behind DNSMadeEasy DNS, with no visible CDN/WAF fingerprint (no Cloudflare, no Akamai). The llms.txt returns a 404 — the site has no AI-friendly content map.

Cold-Knowledge Gap

The LLM knows Metro Credit Union as "one of the largest credit unions in Massachusetts" offering standard products, with a "Metro Rewards Checking" account and a community focus. The site itself claims "largest state-chartered credit union in Massachusetts" with 220,000+ members — a specific stat the model does not cite. The model mentions "competitive interest rates" but the site's rates page lists specific products (Mass Save HEAT Loan, Celebration Loan, Prosperity Builder Loan, Citizenship Loan) that the model does not reference. The model knows nothing about the Members Plus Credit Union merger announced April 2026, the 100-year celebration (founded 1926), or the Metro Contigo Spanish-language initiative — all major brand signals present on the live site.

Schema Posture

The homepage and all subpages carry a BankOrCreditUnion schema with full NAP (name, address, phone, email), social profiles, and logo. The mortgages page adds a FAQPage schema with 7 Q&A pairs — the only product page with structured FAQ markup. However, the rates page, checking page, and blog have no product-level schema (Product, LoanOrCredit, BankAccount, Article). The blog uses WebPage rather than Article or BlogPosting schema, which weakens its eligibility for AI-generated answer citations. The WebPage schema on every page includes a SearchAction with potentialAction — a strong but underutilized signal.

External Signals

The site has a full Spanish-language sub-site at /contigo/ with its own sitemap indexed in robots.txt — a bilingual AI-optimization asset. The blog's top story (April 2026) announces a merger with Members Plus Credit Union, a material event the model knows nothing about cold. The site claims "Voted #1 Credit Union in Best of 2025 Awards" by Banker & Tradesman, but no external press or Reddit threads surfaced in search results — the award citation lacks off-domain corroboration that AI engines would use for confidence scoring. The sameAs array links to Twitter/X, Facebook, LinkedIn, Instagram, YouTube, and Vimeo, but no Google Business Profile or Yelp URL is included.

Findings

1. No AI-bot rules in robots.txt Medium

   The robots.txt file contains no directives for AI crawlers, leaving them unrestricted but also unmanaged, which poses a risk of unmonitored scraping.

   What to change: Add specific rules for AI bots such as GPTBot and ClaudeBot to control access and prevent potential abuse.

2. Missing llms.txt file Medium

   The site returns a 404 for llms.txt, meaning there is no AI-friendly content map to guide language models to key pages.

   What to change: Create an llms.txt file listing important pages like rates, products, and the merger announcement.

3. LLM unaware of Members Plus merger High

   The model's cold knowledge does not include the April 2026 merger with Members Plus Credit Union, a major brand event that should be surfaced in AI answers.

   What to change: Publish a dedicated press release page with structured data and ensure it is indexed and linked prominently.

4. Product pages lack structured schema High

   The rates page, checking page, and blog use only WebPage schema instead of Product, LoanOrCredit, BankAccount, or Article schema, reducing their eligibility for AI citations.

   What to change: Add LoanOrCredit schema to rates pages, BankAccount schema to checking pages, and Article schema to blog posts.

5. Blog uses WebPage instead of Article schema Medium

   The blog posts are marked up as WebPage rather than Article or BlogPosting, which weakens their chance of being used in AI-generated answers.

   What to change: Implement Article or BlogPosting schema on all blog entries.

6. Award claim lacks external corroboration Medium

   The site claims 'Voted #1 Credit Union in Best of 2025 Awards' but no external press or review sites mention this award, reducing its credibility for AI citation.

   What to change: Seek coverage from local news or industry publications and link to those articles from the site.

7. Google Business Profile missing from sameAs Low

   The sameAs array includes social media but not Google Business Profile or Yelp, which are important for local AI visibility.

   What to change: Add the Google Business Profile and Yelp URLs to the sameAs array in the schema.

8. LLM unaware of Spanish-language sub-site Medium

   The model's cold knowledge does not mention the Metro Contigo Spanish-language initiative, which is a significant brand asset for bilingual AI queries.

   What to change: Ensure the Spanish sub-site is linked from the homepage and included in the sitemap with hreflang annotations.

9. LLM unaware of 100-year celebration Low

   The model does not know about the credit union's 100-year anniversary (founded 1926), a key trust signal that should be surfaced.

   What to change: Prominently feature the centennial on the homepage and in schema markup.

What's working

• All AI crawlers receive full HTML content — Every major AI crawler gets a 200 response with identical HTML to the browser baseline, ensuring complete content visibility.
• BankOrCreditUnion schema with full NAP — All pages include BankOrCreditUnion schema with name, address, phone, email, social profiles, and logo, providing strong entity identification.
• FAQPage schema on mortgages page — The mortgages page includes FAQPage schema with 7 Q&A pairs, which can be used directly in AI answers.
• Full Spanish-language sub-site at /contigo/ — The site has a complete Spanish version with its own sitemap, serving bilingual audiences and improving AI visibility for Spanish queries.
• SearchAction schema on every page — Every page includes a SearchAction with potentialAction, enabling AI systems to understand site search functionality.
• Blog covers major merger event — The blog announces the Members Plus Credit Union merger, providing timely content that can be cited by AI if properly structured.