Public sample report · We ran it on ourselves

A real report on visdenda.com, checked 18 November 2026

Every number below came from the live site on 18 November 2026: eight pages fetched, all 206 sitemap URLs checked, robots.txt and llms.txt parsed. Where the site is clean the table says pass; the overruns and gaps are in the queue with an acceptance line.

Last reviewed 18 November 2026

The numbers, before the detail

Six counters, all collected in one pass. The sitemap sweep is exhaustive; the page-level checks read eight URLs across home, pricing, features, blog, tools and directories.

Sitemap URLs
206
All 206 returned HTTP 200
Sampled pages
8/8
HTTP 200 · self-canonical · one H1
AI crawler fetch access
4/4
robots.txt allows all four on public paths; ai-train=no signalled
llms.txt coverage
42/206
Sitemap URLs mentioned; no /blog URL appears
Structured data
8/8
Sampled pages with JSON-LD that parses
Snippet overruns
3
2 titles over 60 · 1 description over 160

Eight pages, read the way a crawler reads them

One row per sampled URL: the status a fetch gets, the title tag length Google has to fit, the canonical target, the meta robots directive, the JSON-LD types served, and whether llms.txt points an agent at the page.

Sampled pages with status, title length, canonical, robots, structured data and llms.txt listing, checked 18 November 2026
PageStatusTitleCanonicalRobotsStructured datallms.txtVerdict
/20053 chSelfNoneOrganization · SoftwareApplication · WebSiteYes
Pass
/pricing20049 chSelfNoneBreadcrumbList · Organization · ProductYes
Pass
/features/search-performance20055 chSelfNoneBreadcrumbList · TechArticleYes
Pass
/blog/verify-an-seo-fix20062 chSelfNoneBreadcrumbList · TechArticleNo
Title 62

Trim 2 characters

/tools/page-seo-check20054 chSelfindexBreadcrumbList · Organization · TechArticleNo
Pass
/checkers20052 chSelfNoneBreadcrumbList · CollectionPage · OrganizationYes
Pass
/crawlers20031 chSelfNoneBreadcrumbList · CollectionPage · Dataset · FAQPage · ItemList · OrganizationYes
Pass
/blog/alternatives20066 chSelfNoneBreadcrumbList · OrganizationNo
Title 66 · Desc 197

Trim both to the window

Checked 18 November 2026 from 03:26 UTC · lengths after entity decoding · canonical “Self” means the href equals the final URL.

What the pass turned up

Four fixes and one standing check from the same pass, ordered by how quickly a reader meets the consequence. Each row states the evidence, why it matters, and the line we will accept as done.

  1. 01

    Two titles and one description run past the snippet window

    Finding
    /blog/alternatives ships a 66-character title and a 197-character description; /blog/verify-an-seo-fix ships a 62-character title.
    Why it matters
    Google truncates by pixel width and rewrites long snippets, so the pitch written for the page is not the one a searcher reads.

    Done when Every sampled title is 60 characters or fewer and every description 160 or fewer, re-read in the SERP preview.

    SERP preview →
  2. 02

    llms.txt leaves most of the site, and every post, out

    Finding
    llms.txt names 51 unique URL strings and mentions 42 of the 206 sitemap URLs. No /blog URL appears, including the comparisons hub.
    Why it matters
    An agent that reads llms.txt gets the docs and the hubs, but not the posts and comparisons that answer commercial questions.

    Done when llms.txt lists /blog and every published post, or the omission is written policy; the llms.txt checker reports no missing page links.

    llms.txt checker →
  3. 03

    The comparison hub has no page or list type

    Finding
    /blog/alternatives serves only BreadcrumbList and Organization JSON-LD; neighboring templates declare TechArticle, CollectionPage, Dataset or Product.
    Why it matters
    A hub with no page or list type leaves its structure to inference while every neighboring page tells machines exactly what it is.

    Done when /blog/alternatives emits CollectionPage or an ItemList of the comparisons and validates with no errors.

    Structured data checker →
  4. 04

    The home and pricing pages declare no Cache-Control

    Finding
    / and /pricing return no Cache-Control header; the six sampled inner pages return public, max-age=0, must-revalidate.
    Why it matters
    The edge is already caching both (cf-cache-status HIT on / with Age 11,973 s; EXPIRED on /pricing), but browsers and downstream caches get no declared rule.

    Done when Both routes return public, max-age=0, must-revalidate on a cache-busted request, matching the inner pages.

    Read from the response headers; no dedicated checker yet.

  5. 05

    Crawler fetch is allowed while training is declined

    Finding
    robots.txt allows GPTBot, ClaudeBot, PerplexityBot and Google-Extended on public paths, then Content-Signal declares ai-train=no, search=yes, ai-input=no.
    Why it matters
    Fetch permission and training permission are separate consents. The split is deliberate, but an edge rule can narrow the first without anyone noticing.

    Done when After any edge or WAF change, robots.txt still allows the four named bots on public paths and Content-Signal still declares ai-train=no.

    AI crawler access →

Locked in the public sample

The rows only your own connection can fill

These four rows are not missing because the audit skipped them; no public crawl can read them. They are private to a property, an account or a competitor list, and they appear beside the public checks once you connect your own site.

  • Search Console clicks

    Requires connection

    Clicks, impressions, CTR and average position per URL and query for the property you connect.

    Private to your Search Console property

    Connect Search Console →
  • GA4 sessions

    Requires connection

    Sessions, engaged sessions and conversions per landing page.

    Private to your GA4 property or first-party beacon

    GA4 setup →
  • Tracked keywords

    Requires connection

    The terms you add, with volume and difficulty. No public row exists.

    Private to your account

    Keyword tracking →
  • Competitor changes

    Requires connection

    New, removed and changed competitor URLs from public sitemaps, against a list you configure.

    Private to your competitor list

    Competitor monitoring →

How the checks were run, and what they miss

One collection pass, one public surface. The method is deliberately narrow so every number on this page can be reproduced with curl.

  1. 01 · Fetch

    Eight URLs fetched with curl on 18 November 2026 from 03:26 UTC: /, /pricing and six URLs taken from the live sitemap.

  2. 02 · Parse

    Each HTML response was parsed for title, meta description, canonical, meta robots, H1 count and JSON-LD @type values. Lengths count characters after entity decoding.

  3. 03 · Sweep

    Every one of the 206 <loc> entries in sitemap.xml was requested (GET, 20 s timeout, six parallel workers) and the status recorded.

  4. 04 · Rules

    robots.txt was read as user-agent groups; the four named AI crawlers were checked against the paths they would fetch.

  5. 05 · Index

    Every absolute URL in llms.txt was extracted and intersected with the sitemap URL set.

  6. 06 · Headers

    Response headers were recorded for / and /pricing on a cache-busted request, plus every sampled page.

What this page does not show

Search Console clicks, GA4 sessions, tracked keyword rows and competitor changes are private to a connected property, an analytics account, an account keyword list or a configured competitor list. They are not part of a public surface check; the public sample stops at what anyone can fetch.

  • Index status, impressions and rankings: whether Google indexed a URL and what it earned in Search Console. That needs a connected property, not a public fetch.
  • Traffic, sessions, conversions and revenue: GA4 or the first-party beacon.
  • Keyword demand and competitor movement: tracked terms, volumes, difficulty and competitor page changes are account data.
  • JavaScript-rendered output, Core Web Vitals, backlinks, local packs or mobile rendering. This sample reads the no-JS HTML response only.
  • Anything after the 18 November 2026 collection pass. A deploy can change every row; this is a point-in-time check.

Audit trail

Raw evidence

The commands, response headers and per-page values behind every row are kept in the repository's SEO records.

docs/seo/raw/me-report-2026-11-18.json

Checked 18 November 2026, from 03:26 UTC. Re-running the commands after a deploy is the only way to refresh the table.

The product-wide method behind these checks — sources, windows, joins and limits — is on the methodology page.

Run it on your site

Get the same report for the URL you own

Free scans: 3 a day, no login. Connect Search Console to fill the private rows.