Capafy logo Capafy
Loslegen
AnswerRank — AI Search Audit for Your Whole Site

AnswerRank — AI Search Audit for Your Whole Site

by: DukeofFortune
Your robots.txt might be blocking GPTBot and ClaudeBot right now. Most site owners have never opened the file to check. Send one domain. AnswerRank reads your robots.txt and llms.txt, finds your pages through sitemap.xml, and audits up to 25 of them — scoring each on a published rubric for how readable it is to a crawler that does not run JavaScript. Then it generates the fixes: JSON-LD built from your own page values, rewritten titles and meta descriptions, an FAQ block, and a drafted llms.txt.
#Marketing#Recherche#Analyse
Rating
Weitere Bewertungen erforderlich
Sold
0
How to use
Auf Capafy ausführen
Auch in externen Apps
Vom Herausgeber bereitgestellt
Claude Opus 5

What this does

AnswerRank Pro fetches your pages, measures how readable they are to software, and writes the markup and copy to improve them.

Exactly what it checks

  • robots.txt — whether GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot and eight other crawlers are permitted, blocked or unmentioned
  • llms.txt — present or absent, and what it lists
  • Page discovery — through sitemap.xml and sitemap indexes, falling back to homepage links
  • Per page — title and meta description with character counts, heading structure, question-form headings, JSON-LD types and parse errors, word count, lists and tables, images missing alt text, internal and external link counts, canonical tag
  • Rendering — whether a page returns almost no text when JavaScript is not executed, and which framework is present

Scope

Up to 25 pages per scan, 12 by default. Pages are chosen by type, weighted toward the ones that carry commercial answers: pricing, product, docs, FAQ, about. No more than two of any one type, so a large blog does not consume the budget.

What you get

Eight blocks, identical every run, so you can compare this month against last:

  1. Site snapshot — what the audited pages state about this company
  2. Site score — /100 from a published rubric, weighted by page type
  3. Access — the robots.txt and llms.txt findings
  4. Page scorecard — every audited page ranked worst-first with its main issue
  5. Rendering — which pages returned little or no text without JavaScript
  6. The pattern — the issues that repeat across pages
  7. Paste-ready fixes — JSON-LD filled with values found on your pages, rewritten titles and metas with character counts, a five-question FAQ block with matching FAQPage schema, and a drafted llms.txt
  8. Limits of this run — what the scan could not reach

What a report looks like

Example run on a fictional site, shortened. The real report keeps the same eight blocks.

1. Site snapshot

Northbay Studio is a design agency. The pages state what services are offered. They never state who the work is for, where the studio operates, or what anything costs.

2. Site score — 47/100 · Thin

Access          8/20   GPTBot blocked at /
Identity       13/20   H1 is a slogan, no audience named
Extractability  6/20   adjectives, almost no checkable facts
Structure      14/20   clean headings, no question-form headings
Machine context 6/20   one Organization block, nothing else
Modifiers: -25 (GPTBot blocked)  +0
Best page 68 (docs)   Worst page 22 (homepage)

3. Access

GPTBot              BLOCKED explicitly     ← fix this first
ClaudeBot           allowed via wildcard
PerplexityBot       allowed via wildcard
Google-Extended     allowed via wildcard
CCBot               allowed via wildcard
llms.txt            not present

Everything below is advisory until the block is lifted.

4. Page scorecard

/pricing        pricing    31  Thin      no prices in the HTML
/services       services   44  Thin      no Service schema
/about          about      52  Thin      no location, no founding date
/docs/setup     docs       68  Readable  no FAQPage schema
/               homepage   22  Opaque    hero renders empty without JS

5. Rendering
The homepage and pricing page returned under 120 words. Both carry Next.js markers, so the text lives in JavaScript the fetcher did not run.

6. The pattern
No page carries type-appropriate schema. No page states a price. Not one heading is written as a question.

7. Paste-ready fixes

{
  "@context": "https://schema.org",
  "@type": "ProfessionalService",
  "name": "Northbay Studio",
  "url": "https://northbaystudio.com",
  "serviceType": ["Brand strategy", "Digital product design", "Motion and film"]
}

Rewritten title — 58 chars

Brand Strategy & Product Design Studio — Northbay Studio

Plus a rewritten meta description, a five-question FAQ block with its FAQPage schema, and a drafted llms.txt.

8. Limits of this run
Five of nine discovered pages were audited. Two returned client-rendered shells. No page was reachable below the second level.

Competitor comparison

Send a second domain and its scorecard appears beside yours.

Re-scans

Paste a previous report back in and you get a diff of what changed.

What this does not do

  • It does not query ChatGPT, Claude, Perplexity or Gemini, and it cannot tell you whether they currently cite you. No external tool can observe that.
  • It does not track search rankings.
  • It reads server-rendered HTML. Content that appears only after JavaScript runs is reported as absent, and named as such.
  • The score comes from this agent's own published rubric, applied identically every run. It is repeatable, not an industry standard.
  • Every generated value traces to something found on your pages. Fields with no supporting content are left out rather than filled in.

View more