exactship.com

robots.txt exists but contains no parseable directives. Check for HTML being served in its place.

{
  "bytes": 425593,
  "status": 200,
  "directives": 0
}

Fix: sitemap parsed to zero URLs; not referenced from robots.txt. A sitemap referenced from robots.txt is how a crawler discovers pages that are not linked from the homepage.

{
  "exists": true,
  "urlCount": 0,
  "referencedInRobots": false
}

llms.txt exists but only 0 links carry a description. Add a short description after each link — an undescribed link list gives a model no basis for choosing what to fetch.

{
  "present": true,
  "anyLinks": 0,
  "llmsFullTxt": true,
  "describedLinks": 0
}

Your robots.txt welcomes these agents, but ClaudeBot received 429 — the block is happening at your CDN or WAF, before robots.txt is ever read. Cloudflare in particular blocks AI crawlers by default on every plan. One setting change, and it is invisible until someone tests it.

{
  "probes": [
    {
      "name": "GPTBot",
      "status": 200,
      "ttfbMs": 634
    },
    {
      "name": "ClaudeBot",
      "status": 429,
      "ttfbMs": 237
    },
    {
      "name": "PerplexityBot",
      "status": 200,
      "ttfbMs": 163
    }
  ],
  "medianTtfbMs": 237,
  "edgeBlockingDespiteRobots": true
}

Detected: Cookie wall. An agent cannot dismiss a consent dialog or solve a challenge — whatever sits behind it is unreachable. Cookieless analytics removes the need for a banner entirely.

{
  "markers": [
    "Cookie wall"
  ]
}

No headless render was captured, so parity could not be measured. Re-run with rendering enabled.

{
  "rendered": false
}

Missing: type-specific markup (Product, Offer). For a ecommerce site, Product or Offer is what lets an agent understand what you actually offer rather than that you exist.

{
  "matched": [],
  "missing": [
    "type-specific markup (Product, Offer)"
  ],
  "siteType": "ecommerce",
  "typesFound": [
    "BreadcrumbList",
    "EntryPoint",
    "ItemList",
    "ListItem",
    "PropertyValueSpecification",
    "ReadAction",
    "SearchAction",
    "Thing",
    "WebPage",
    "WebSite"
  ]
}

Headings and landmarks are how a machine builds an outline of the page. One h1, no skipped levels, and real main/nav/footer elements — all template-level fixes, applied once.

{
  "problems": [
    "https://exactship.com: skipped heading level",
    "https://exactship.com/pricing/: skipped heading level",
    "https://exactship.com/products/: skipped heading level",
    "https://exactship.com/about-us/: skipped heading level",
    "https://exactship.com/company/events/: skipped heading level",
    "https://exactship.com/company/careers/: skipped heading level"
  ],
  "singleH1": 6,
  "allLandmarks": 6,
  "pagesCrawled": 6,
  "noSkippedLevels": 0
}

Fix: 6 title(s) outside 15–70 characters; duplicate titles across pages.

{
  "goodTitles": 0,
  "pagesCrawled": 6,
  "uniqueTitles": false,
  "goodDescriptions": 6,
  "completeOpenGraph": 6
}

Visible text is 6.7% of your markup. Deeply nested wrapper elements make a page expensive to parse and dilute the content an agent extracts. Script and style contents are already excluded from this measure, so this is markup weight, not framework overhead.

{
  "perPage": [
    0.067,
    0.067,
    0.067,
    0.067,
    0.067,
    0.067
  ],
  "averageRatio": 0.067
}

660 of 894 content images have no alt text. An agent cannot see the image; the alt attribute is the only description it gets.

{
  "withAlt": 234,
  "coverage": 0.26,
  "contentImages": 894
}

Fix: 5 page(s) missing or with a mismatched rel=canonical. Duplicate hosts split whatever authority the page has earned.

{
  "hostVariants": [
    {
      "url": "https://www.exactship.com",
      "status": 301,
      "location": "https://sellercloud.com"
    },
    {
      "url": "http://exactship.com",
      "status": 301,
      "location": "https://sellercloud.com"
    }
  ],
  "pagesCrawled": 6,
  "selfConsistentCanonicals": 1
}

No forms were found on the crawled pages, so form usability could not be assessed. If you add a contact or signup form, give every input a real label and a correct input type.

{
  "formsFound": 0
}

Fix: 40% of links use contextless text like "click here".

{
  "anchors": 624,
  "genericLinks": 247,
  "genericRatio": 0.4,
  "hasBreadcrumb": true,
  "clickableNonAnchors": 0
}

No machine-readable endpoint of any kind. Most sites score zero here, which is exactly why it is the cheapest differentiation on this rubric: an agents.json is a static file, and a documented read-only API is usually a subset of something you already run.

{
  "found": []
}

Fix: no Offer schema carrying price and availability; cart or checkout path did not return 200 without login; no guest or express checkout signal detected. Agentic commerce depends on an agent being able to read a price and reach a checkout without an account.

{
  "offerSchema": false,
  "cartReachable": false,
  "expressCheckout": false,
  "productPageFound": true
}

Search exists but posts rather than using a GET query parameter, so an agent cannot construct a results URL directly. Switching to GET is usually a one-line change, and advertising it via WebSite SearchAction schema makes it discoverable.

{
  "usesGet": false,
  "searchForm": false,
  "searchAction": true
}

Fix: 1 insecure subresource reference(s).

{
  "hsts": true,
  "tlsValid": true,
  "mixedContentReferences": 1
}

Fix: no Organization schema. sameAs links to a company profile are how a model corroborates that you are who the page says you are.

{
  "sameAsCount": 0,
  "contactDiscoverable": true,
  "hasOrganizationSchema": false
}

Fix: sitemap has no lastmod values; only 0% of sitemap URLs modified in the last 90 days. Staleness is a ranking signal for models deciding what to cite.

{
  "sitemapUrls": 0,
  "withLastmod": 0,
  "visibleDates": true,
  "modifiedLast90Days": 0
}

security.txt is present but invalid: no Contact field; no Expires field. An expired Expires field makes the whole file non-conformant, so this fails even though the file was correct when it was written.

{
  "valid": false,
  "hasContact": false
}