Monsieur Parfait

Can AI engines actually read your site?

• Fix

  Sitemap.xml present and submitted

  A complete sitemap.xml exists, lists all important pages, is referenced in robots.txt, and is submitted to Google Search Console and Bing Webmaster Tools.

  No sitemap found at /sitemap.xml and no Sitemap: directive in robots.txt.

  What to do: Generate a sitemap.xml listing all important pages and submit it to Google Search Console and Bing Webmaster Tools.

• Improve

  Canonical tags set correctly

  Every page declares a canonical URL, with no conflicting or self-referential errors. Critical for e-commerce with faceted navigation.

  No <link rel='canonical'> tag found on the homepage.

  What to do: Add a canonical tag pointing to the preferred URL of each page. Critical when you have URL parameters or faceted navigation.

• Manual review

  Cloudflare WAF doesn't block AI bots

  Cloudflare's 'Block AI Scrapers and Crawlers' setting and default Bot Fight Mode can silently block AI crawlers. Verify the rules don't apply to GPTBot, ClaudeBot, etc.

  Cloudflare is not detected in front of this site, so the 'Block AI Scrapers' setting doesn't apply here.

• Manual review

  Critical content is not PDF-only

  Case studies, product specs, and white papers exist as crawlable HTML pages, not just downloadable PDFs. AI cites HTML pages roughly 10x more often than PDFs.

  Multi-page crawl needed to detect PDF-only content.

  What to do: Ensure case studies, product specs, and white papers exist as crawlable HTML, not only PDFs. AI cites HTML pages 10x more than PDFs.

• Good

  Content rendered server-side

  Your important content is in the initial HTML response, not rendered by JavaScript afterwards. Most AI crawlers don't execute JS, so client-rendered content is invisible to them.

  Homepage HTML contains ~331 words of visible text — AI crawlers can read it without JavaScript execution.

• Good

  Full HTTPS, no mixed content

  Every page loads over HTTPS with a valid certificate, and no internal links or assets fall back to HTTP.

  Site loads over HTTPS with no detected mixed content.

• Good

  robots.txt allows AI crawlers

  Your robots.txt explicitly allows GPTBot, ClaudeBot, PerplexityBot, CCBot, and Google-Extended. Blocking any of them — even by accident — makes you invisible to that platform.

  robots.txt does not block any known AI crawler.

Can AI agents not just read, but act on your site?

• Fix

  Agent Skills index published

  A /.well-known/agent-skills/index.json file (Agent Skills Discovery RFC v0.2.0) lists callable skills agents can invoke. Each skill has a name, type, description, URL and sha256 digest. Cloudflare-led standard, increasingly required.

  No Agent Skills index found at /.well-known/agent-skills/index.json.

  What to do: Publish an Agent Skills index (Cloudflare's RFC v0.2.0) listing callable skills agents can invoke. Each skill needs a name, type, description, URL and sha256 digest.

• Fix

  MCP Server Card available

  A /.well-known/mcp.json file declares which actions agents can take on your site (book, submit, query). Emerging standard backed by Anthropic and adopted by Claude Desktop and ChatGPT. Fewer than 100 sites worldwide have one.

  No MCP server card found at /.well-known/mcp.json.

  What to do: Publish an MCP server card declaring actions agents can take on your site. Emerging standard backed by Anthropic. Fewer than 100 sites worldwide have one — strong differentiator.

• Fix

  Markdown content negotiation supported

  When an AI agent sends 'Accept: text/markdown', your server responds with a clean Markdown version of the page. Saves tokens for the agent — agents are more likely to cite low-token sources. Less than 4% of sites do this.

  Asked for markdown but got Content-Type: text/html.

  What to do: Set up your server to return text/markdown when requested. Either pre-generate .md twins of your key pages or run a lightweight HTML→Markdown conversion middleware.

• Fix

  RFC 8288 Link headers for discovery

  HTTP Link response headers point agents to your well-known resources (mcp-server, api-catalog, agent-skills, sitemap, llms.txt) without parsing HTML. Agents using HEAD requests or Markdown negotiation only see headers, not <link> tags.

  No HTTP Link headers found on the homepage.

  What to do: Add a Link response header pointing to your well-known resources (llms.txt, sitemap.xml, mcp.json, etc). Agents using HEAD requests or markdown negotiation never see <link> tags in HTML — only headers.

• Fix

  llms.txt file published

  An llms.txt at the site root tells AI agents what your site is about and points to the most authoritative pages. Less than 1% of sites have one — a free differentiator.

  No llms.txt found at /llms.txt.

  What to do: Publish an llms.txt at the site root summarising what the site is about and pointing AI agents to your most authoritative pages. See llmstxt.org for the format. Fewer than 1% of sites have one.

• Improve

  Content Signals declared in robots.txt

  Use the Content-Signal directive (format: search=yes, ai-input=yes, ai-train=no) to declare how AI may use your content. Distinct from Allow/Disallow — silence is increasingly interpreted as restrictive.

  robots.txt exists but doesn't declare Content-Signal directives.

  What to do: Add 'Content-Signal: search=yes, ai-input=yes, ai-train=no' (adjust to your preference) under your User-agent blocks. Silence is increasingly interpreted as restrictive by AI platforms.

• Manual review

  Forms readable and submittable by agents

  Forms use semantic HTML (proper labels, autocomplete attributes, aria-required), don't rely on heavy JS for validation, and accept standard form-encoded POST. An agent can fill them without computer-vision tricks.

  No <form> found on the homepage.

  What to do: If you have forms on other pages (contact, signup, search), ensure they use semantic HTML — proper labels, autocomplete attributes, and standard form-encoded POST. Agents struggle with custom JS form components.

• Manual review

  OAuth Protected Resource Metadata

  If you offer protected APIs, publish /.well-known/oauth-protected-resource (RFC 9728) declaring the authorization servers that can issue tokens for it and the supported scopes. Lets agents discover how to authenticate before attempting access.

  Manual review required — only relevant if you offer protected APIs.

  What to do: If you offer protected APIs, publish /.well-known/oauth-protected-resource (RFC 9728) declaring authorization servers and supported scopes.

• Manual review

  Public API catalog (RFC 9727)

  /.well-known/api-catalog (application/linkset+json) lists your public APIs with service-desc (OpenAPI), service-doc and status endpoints. Agents discover and integrate without scraping.

  No API catalog found at /.well-known/api-catalog. Only relevant if you offer public APIs.

  What to do: If you offer public APIs, publish /.well-known/api-catalog (RFC 9727) listing them with service-desc (OpenAPI), service-doc and status endpoints.

How clearly machines can understand who you are.

• Fix

  FAQ schema on key pages

  FAQ schema markup on pages that genuinely answer customer questions. Don't fake it — Google penalises FAQ stuffing, and AI engines verify against page content.

  No FAQPage schema found on the homepage.

  What to do: If you answer common customer questions anywhere on your site, mark them up with FAQPage JSON-LD. Don't fake it — Google penalises FAQ stuffing.

• Fix

  LocalBusiness schema

  LocalBusiness (or specific subtype like Restaurant, Store) schema with address, hours, geo coordinates, price range, and aggregated review rating.

  No LocalBusiness schema (or subtype) found.

  What to do: If you have a physical location, add LocalBusiness JSON-LD with address, geo coordinates, hours, priceRange, and aggregateRating. Use a specific subtype (Restaurant, Store, Dentist, etc) when applicable.

• Fix

  Organization schema

  JSON-LD Organization schema on the homepage with name, logo, URL, sameAs links to social profiles, address, and contact info. This is the single most-cited structured data block by AI.

  No JSON-LD Organization schema found on the homepage.

  What to do: Add a JSON-LD Organization block on the homepage with name, logo, URL, sameAs (social links), address, and contact info. This is the most-cited structured data block by AI.

• Fix

  sameAs linking to social and directories

  Organization schema includes sameAs URLs to your LinkedIn, Wikipedia, Crunchbase, industry directories — all the entities AI uses to build a trustworthy picture of you.

  No sameAs URLs found in any JSON-LD block.

  What to do: Add sameAs URLs to your Organization schema pointing to your LinkedIn, Wikipedia, Crunchbase, and industry directories. AI uses these to build a trustworthy picture of your entity.

• Manual review

  Article schema on content pages

  Blog posts and case studies use Article schema with author, datePublished, and dateModified. AI weighs freshness signals.

  No Article schema on the homepage. Typically lives on blog post and case study pages.

  What to do: Blog posts and case studies should use Article (or BlogPosting) schema with author, datePublished, and dateModified.

• Manual review

  BreadcrumbList schema

  BreadcrumbList schema on category and product pages helps AI understand your site hierarchy.

  No BreadcrumbList schema on the homepage. More relevant on deep pages.

  What to do: Add BreadcrumbList JSON-LD on category and product pages to help AI understand your site hierarchy.

• Manual review

  Person schema for leadership

  Founders and key staff have Person schema with role, credentials, and sameAs links to LinkedIn. Helps AI characterize who runs the business.

  No Person schema for leadership on the homepage.

  What to do: Founders and key staff should have Person schema with role, credentials, and sameAs links to LinkedIn. Typically on the About or Team pages.

• Manual review

  Product schema with reviews

  Product schema on every product page with brand, GTIN/SKU, price, availability, and aggregated reviews. Drives both rich results in Google and citations in shopping AI agents.

  No Product schema on the homepage. Only relevant for e-commerce.

  What to do: If you sell products, ensure each product page has Product JSON-LD with brand, GTIN/SKU, price, availability, and aggregateRating. The homepage typically doesn't need it.

• Manual review

  Service schema per core offering

  Each main service has its own page with Service schema, describing what it is, who it's for, and areas served.

  No Service schema on the homepage. Typically lives on individual service pages.

  What to do: Each main service should have its own page with Service JSON-LD describing what it is, who it's for, and areas served.

How easy it is for AI to quote you.

• Fix

  Long-form About page

  Substantive About page (1,000+ words) covering history, philosophy, team, and concrete differentiators. The single page AI cites most when answering 'who is X' queries.

  Homepage has only ~331 words — likely too thin for AI to characterize the business.

  What to do: Build a long-form About page (1,000+ words). The single page AI cites most when answering 'who is X' queries.

• Manual review

  Case studies in crawlable HTML

  Each case study is a public HTML page with client name (where permitted), problem, solution, technologies used, and measurable outcomes. Not just a PDF download.

  Case studies presence and quality need a multi-page crawl.

  What to do: Each case study should be a public HTML page with client, problem, solution, technologies, and measurable outcomes.

• Manual review

  Industry-vertical pages

  Dedicated pages for each vertical you serve, with the language, use cases, and outcomes specific to that industry. AI excels at recommending vertical-specific service providers.

  Industry-vertical pages need a multi-page crawl to evaluate.

  What to do: Build dedicated pages for each vertical you serve, with industry-specific language and use cases.

• Manual review

  Technology / platform pages

  Explicit pages naming every major technology you work with (e.g., Siemens, Rockwell, Shopify, Salesforce). AI looks for exact-match technology terms when matching buyer intent.

  Technology/platform pages need a multi-page crawl to evaluate.

  What to do: Build explicit pages naming every major technology you work with (Siemens, Salesforce, etc).

Whether AI can describe who you are in two sentences.

• Manual review

  Consistent NAP (name, address, phone)

  NAP information matches exactly across every directory and platform. A single typo in your address can cost local visibility.

  NAP consistency requires cross-referencing your site with directories.

  What to do: Verify your name, address, and phone number match exactly on website, Google Business, LinkedIn, social profiles, and directories.

• Manual review

  Wikipedia mention or page

  A Wikipedia page (where notability allows) or mentions in relevant Wikipedia articles. Wikipedia is one of the most heavily weighted citation sources by AI engines.

  Wikipedia presence requires manual research.

  What to do: If notability allows, get a Wikipedia page or relevant mentions. Wikipedia is one of the most heavily weighted citation sources by AI.

What outside sources back you up.

• Manual review

  Industry directory listings

  Listings in directories specific to your sector, with complete profiles and consistent information.

  Off-site signal — requires manual research or external API.

  What to do: List in directories specific to your sector with complete, consistent profiles.

• Manual review

  Original research or data publication

  Published reports based on data only you have (industry surveys, methodology papers, internal benchmarks). The highest-leverage long-game move — gets cited for years.

  Off-site signal — requires manual research or external API.

  What to do: Publish reports based on data only you have. Highest-leverage long-game move — gets cited for years.

• Manual review

  Speaking engagements and panel slots

  Conference talks, panels, and podcast appearances. Each one produces durable content that AI later cites.

  Off-site signal — requires manual research or external API.

  What to do: Speak at conferences, on panels, podcast appearances. Each produces durable content AI later cites.

• Manual review

  Third-party reviews on trusted sites

  Reviews on Google, TripAdvisor, Trustpilot, G2, Capterra, or sector-specific directories. AI weighs aggregated review presence as a credibility signal.

  Off-site signal — requires manual research or external API.

  What to do: Build review presence on Google, TripAdvisor, Trustpilot, G2, Capterra, or sector-specific directories.

Visibility for queries with a place attached.

• Manual review

  City and neighborhood named on the site

  Your city and neighborhood appear naturally in the homepage copy, About, and footer. AI cross-references these against the place names in user queries.

  Local signal — requires Google Business Profile API or manual verification.

  What to do: Make sure your city and neighborhood appear naturally in homepage copy, About, and footer.

• Manual review

  Google Business & maps presence

  Your presence across Google Business Profile, Google reviews, and the wider maps ecosystem (Apple Maps, Bing Places).

  • Manual review Google Business Profile fully filled

    Local signal — requires Google Business Profile API or manual verification.

    What to do: Fill every field of your Google Business Profile — categories, services, attributes, photos, hours, posts.

  • Manual review Recent photos on Google Business

    Local signal — requires Google Business Profile API or manual verification.

    What to do: Upload fresh photos monthly: interior, products, team, events. Drives ranking and click-through.

  • Manual review Google reviews — volume and recency

    Local signal — requires Google Business Profile API or manual verification.

    What to do: Build a steady inflow of reviews with proprietor responses on most.

  • Manual review Apple Maps and Bing Places presence

    Local signal — requires Google Business Profile API or manual verification.

    What to do: List on Apple Maps and Bing Places, both increasingly cited by AI.

Speed and stability metrics that gate ranking.

• Manual review

  Cumulative Layout Shift (CLS) under 0.1

  Your layout doesn't jump around as it loads. Frequent CLS issues come from images without dimensions and late-loading ads.

  Cumulative Layout Shift requires the Google PageSpeed Insights API — deferred to a later pack.

  What to do: Ensure layout doesn't jump as it loads. Set explicit dimensions on images and reserve space for ads.

• Manual review

  Interaction to Next Paint (INP) under 200ms

  Your site responds instantly to clicks and taps. INP replaced FID as a Core Web Vital in 2024.

  Interaction to Next Paint requires the Google PageSpeed Insights API — deferred to a later pack.

  What to do: Ensure your site responds to clicks under 200ms. Often fixed by reducing JavaScript execution time.

• Manual review

  Largest Contentful Paint (LCP) under 2.5s

  Your main content loads quickly. LCP is one of three Core Web Vitals Google uses as a ranking signal.

  Largest Contentful Paint requires the Google PageSpeed Insights API — deferred to a later pack.

  What to do: Ensure your main content loads under 2.5s. Often fixed by optimizing the hero image and inlining critical CSS.

• Manual review

  Modern image formats and compression

  Images served as WebP or AVIF, properly sized for the device, with lazy loading on below-the-fold content.

  image format and compression requires the Google PageSpeed Insights API — deferred to a later pack.

  What to do: Serve images as WebP or AVIF, properly sized, with lazy loading on below-the-fold content.

• Manual review

  No render-blocking resources

  Critical CSS is inlined. Non-critical CSS and JS are deferred or async-loaded. Especially important for ad-heavy or analytics-heavy sites.

  render-blocking resources requires the Google PageSpeed Insights API — deferred to a later pack.

  What to do: Inline critical CSS; defer non-critical CSS and JS.

Does your site hold up on a phone?

• Manual review

  Readable text without zooming on mobile

  Body text is at least 16px on mobile. Small text triggers Google's mobile usability warnings.

  Mobile font size detection needs computed rendering, deferred to Lighthouse integration.

  What to do: Ensure body text is at least 16px on mobile. Small text triggers Google mobile usability warnings.

• Manual review

  Tap targets large enough on mobile

  Buttons and links are at least 44px and not too close to each other. Failing this hurts both mobile UX and Core Web Vitals.

  Tap target size needs computed rendering, deferred to Lighthouse integration.

  What to do: Ensure buttons and links are at least 44px and not too close to each other.

• Good

  Proper mobile viewport meta tag

  viewport meta tag is set correctly so the site adapts to phone screens. Should be standard, but is still missing on older sites.

  Viewport meta tag is properly set: width=device-width, initial-scale=1.0

Who links to you, and from where.

• Manual review

  Natural anchor text profile

  Backlinks to your site use varied, natural anchor text. Over-optimised anchor text (same keywords repeatedly) triggers spam signals.

  Backlink data — requires Ahrefs/Moz API, deferred to a later pack.

  What to do: Keep backlink anchor text varied and natural; over-optimisation triggers spam signals.

• Manual review

  Strong internal linking structure

  Important pages are linked from many other pages on your site, with descriptive anchor text. Helps AI understand your site hierarchy and what matters most.

  Backlink data — requires Ahrefs/Moz API, deferred to a later pack.

  What to do: Link important pages from many places on your site with descriptive anchor text.