Methodology
A transparent view of the audit pipeline: what we crawl, what we test, how we score, and where the limits are.
We start with the public website. The audit checks HTTP status, canonical signals, sitemap availability, robots.txt directives, llms.txt presence (see note on emerging standards below), important page templates, and whether AI-related crawlers are blocked from key content.
We run category, comparison, problem-aware, and brand-adjacent prompts across ChatGPT, Gemini, and Claude. Prompt sets are designed around buying intent, not vanity keyword volume.
Each response is reviewed for brand mentions, official-domain citations, third-party citations, unsupported claims, competitor mentions, and cases where the AI answer gives no usable source.
The audit reviews schema markup, FAQ entities, organization signals, product specifications, comparison pages, category language, and whether the page answers likely AI-user questions directly.
Competitor visibility is measured against the same prompt set so the report shows whether your issue is technical access, missing category authority, weak comparison content, or stronger external citations around competitors.
Scores combine crawlability, citation behavior, content clarity, structured signals, and competitive context. Fixes are prioritized by expected visibility impact and implementation effort.
We check for llms.txt presence and report it, but we do not weight it heavily in scoring. Google has stated llms.txt provides no SEO benefit today, and no major LLM provider has publicly confirmed they fetch or use llms.txt files. We track this signal because the standard may become meaningful — but we don't oversell it.
AI answers vary by model version, geography, account state, prompt wording, and time.
AnswerAtlas does not have privileged access to OpenAI, Anthropic, Google, Shopify, or their ranking systems.
The audit is a structured visibility snapshot, not a guarantee that any AI system will recommend a brand.
Monitoring exists because AI recommendation visibility can drift after the initial audit.
Start with a free snapshot, then use monitoring if AI recommendation visibility matters continuously.