What does the AI visibility audit check?

It checks robots.txt rules for major AI crawler families, llms.txt files, answer-oriented content and schema, raw HTML facts, Organization identity signals, Wikipedia and Wikidata exact-name results, visible dates, HTTPS, and About and Contact pages.

Does this audit ask ChatGPT or another assistant about my business?

No. Version one is deterministic and makes no paid LLM calls. It measures the public evidence assistants can retrieve. Live recommendation, citation and answer-accuracy prompts are planned for the version two probe layer.

Does adding llms.txt guarantee AI visibility?

No. llms.txt is an emerging convention, not an established ranking or recommendation factor. The audit gives it limited weight and treats normal crawlable pages, accurate schema and corroborating sources as the foundation.

Why do I need an email code?

The six-digit code confirms that the email address belongs to you, replaces a captcha and makes the two-free-reports-per-address limit enforceable. No account or password is created.

* Free · Deterministic v1 · No paid LLM calls

AI visibility

Can AI assistants find, read and recommend your business?

We fetch the public evidence an answer engine can use, from crawler rules and raw HTML to entity links and dated trust signals. You get the score, the evidence and a fix order by effort.

Also free:Website audit Instagram audit Google Ads audit Meta Ads audit Instagram Ads audit AI visibility audit Email deliverability audit Business Profile audit Rank checker

Domain, brand and email

The brand name is used only for exact-name checks in the public Wikipedia and Wikidata APIs. The domain supplies the site evidence.

Public domain

Use the main public website, not a social profile or private dashboard.

Brand name customers use

Use the public trading name, including spacing and punctuation.

Where should we send the report?

We send a verification code, then the private report link. Two free reports per address, shared across Shabang audit tools.

Email me the report, and feel free to follow up once about it. No mailing list and no shared details.

Six digits, ten-minute expiry, no password or account.

Four weighted pillars

The weight is shown before you run the tool. Missing external API data is excluded rather than turned into a zero.

01 · 30%

Access

robots.txt by GPTBot, OAI-SearchBot, ClaudeBot, anthropic-ai, PerplexityBot, Google-Extended, cohere, CCBot and Bytespider. Includes the CCBot and Common Crawl trap, plus llms files.

02 · 30%

Answerability

FAQPage, QAPage and HowTo schema, question headings, address, phone and prices near the top, plus initial HTML word count against script density.

03 · 25%

Entity

Exact brand-name results from Wikipedia and Wikidata, Organization schema completeness, sameAs corroboration and a LinkedIn company link.

04 · 15%

Trust

HTTPS, visible full dates and discoverable About and Contact pages. These make a retrieved recommendation easier to verify and act on.

* Scope

What the score means

It measures public retrieval and verification readiness. It does not claim that an assistant has indexed you, will cite you or will recommend you.

No live assistant prompts in v1

Recommendation presence, answer accuracy and citation consistency need actual assistant responses. Those belong in the v2 probe layer.

llms.txt is emerging

It receives limited weight. It can be a useful summary for compatible tools, but it is not an established ranking factor and should not replace normal pages.

Wikipedia is not a marketing checklist

No article is normal for many valid businesses. Absence is a warning-level context signal, not a failure. Any article or Wikidata item must follow the platform’s sourcing and notability rules.

Prefer a human review?

Thirty minutes, free. We will separate crawler configuration, content, schema and genuine authority work, then give you a practical order.

Book a strategy call WhatsApp us