How do AI agents
see your site?

What this audit actually checks

1. llms.txt manifest — the emerging convention AI crawlers follow for a curated, machine-readable site index. Weight: 15%.
2. llms-full.txt corpus — single-file full-text dump for retrieval- augmented citation. Weight: 8%.
3. AI crawler allow-list in robots.txt — explicit Allow rules for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended. Many sites block by default. Weight: 12%.
4. Schema.org structured data — JSON-LD presence, depth, and graph coverage (Organization, BlogPosting, FAQPage, BreadcrumbList, etc). Weight: 18%.
5. Open Graph + Twitter Card metadata — social-preview metadata completeness for X, LinkedIn, Slack, Discord previewers. Weight: 10%.
6. Speakable schema — SpeakableSpecification on long-form content. Powers Google Assistant + AI Overview voice extraction. Weight: 8%.
7. Heading hierarchy — single h1, structured h2/h3, no level-skips. AI agents use heading structure for content extraction. Weight: 7%.
8. XML sitemap — sitemap presence, URL count, lastmod tags. Weight: 7%.

Why most sites score under 60 in 2026

Most enterprise sites were built before AI agents were a citation surface. They still optimise for Google blue-link SERPs — meta titles, schema, sitemap. AI agents look for additional signals: a curated llms.txtmanifest, an explicit AI-crawler allow-list, Speakable spec on long-form content, and clean structured data on every page rather than only the homepage. The gap between "blue-link ready" and "AI-citation ready" is where most visibility is lost in 2026.

What is AI visibility?

AI visibility is the measurable rate at which AI search engines — ChatGPT Search, Perplexity, Claude Web, Google AI Overviews, Bing Copilot — discover, parse, and cite your domain when users ask brand-relevant or category-relevant questions. It is not the same as AI search ranking, which is the position your domain occupies inside a generated answer. AI visibility is the upstream signal: whether the engine can ingest your site at all, and whether the technical signals on the page make your content eligible for citation. Without visibility, ranking is impossible.

How AI search visibility differs from Google SEO

Traditional Google SEO optimises for blue-link ranking: meta titles, internal links, page authority, backlinks. AI search visibility optimises for citation in generative answers, which depends on a different signal stack —llms.txt manifests, AI crawler allow-lists, structured data depth, Speakable spec on long-form content, and JSON-LD on every page rather than only the homepage. AI agents extract entities and citations differently than Googlebot ranks pages, which is why a site can rank well in Google but be invisible to ChatGPT or Perplexity. The two stacks overlap on schema and sitemap quality but diverge sharply on crawler permissioning and on the llms.txt manifest, which is unique to the AI-search era.

How to improve brand visibility in AI search engines

Short answer: the seven engineering moves below — publish anllms.txt manifest, allow AI crawlers in robots.txt, ship JSON-LD on every page, add SpeakableSpecification to long-form content, keep a semantic heading hierarchy, maintain a clean XML sitemap withlastmod tags, and add a llms-full.txt content corpus. Each move targets a specific signal that ChatGPT Search, Perplexity, Claude Web, and Google AI Overviews use to decide whether your domain is citation-eligible.

The full engineering checklist that moves the audit score most reliably in 2026:

1. Publish an llms.txt manifest at the root of your domain with a curated, machine-readable index of canonical URLs grouped by intent.
2. Allow GPTBot, ClaudeBot, PerplexityBot,Google-Extended, and Applebot-Extended inrobots.txt. Many sites block these by default and lose AI citation eligibility without realising it.
3. Ship JSON-LD on every page: Organization withsameAs links to Crunchbase, LinkedIn, GitHub; BlogPostingon every article; FAQPage on long-form content; BreadcrumbListfor hierarchy.
4. Add SpeakableSpecification to long-form content so Google Assistant and AI Overview voice extraction can pull eligible sentences.
5. Maintain a semantic heading hierarchy — single h1, structuredh2/h3 with no level-skips. AI agents use heading structure for content extraction.
6. Keep a clean XML sitemap with lastmod tags. Stale or missing sitemaps suppress recrawl frequency.
7. Add a llms-full.txt single-file content corpus for retrieval-augmented citation by engines that prefer the bulk format over per-URL crawling.

What strategies improve brand visibility in AI search engines?

Three strategy layers, in order of priority for engineering teams in 2026.

1. Technical-signal coverage — the seven on-page moves listed above ship the AI-citation signal stack: llms.txt, AI crawler allow-list, JSON-LD, Speakable, headings, sitemap, llms-full.txt. Without these, the rest of the strategy does not compound.
2. Entity authority — disambiguate your Organizationentity with sameAs links to Crunchbase, LinkedIn, GitHub, Wikipedia where applicable. AI engines weight cross-referenced entities much more heavily than orphaned ones. Pair this with consistent NAP (name, address, phone) in your structured data and footer.
3. Citation-eligible content depth— write the first 60 words of each long-form page as a self-contained answer to the page's primary question. AI engines extract that opening as a citation candidate. AddFAQPageJSON-LD with 4-6 concrete Q&A pairs per long-form piece, and use 60-120 word answers in the FAQ schema rather than terse one-liners.

The strategy compounds when all three layers are in place. Brands that ship only the technical signal stack improve their crawl-eligibility but still under-cite; brands that ship entity authority without the technical signals never get crawled in the first place. The audit on this page scores the first layer; the second and third layers are content + entity work that follows the audit fixes.

How to improve visibility in Google AI Overviews

Google AI Overviews — the generative summary panel at the top of Google search results — picks citation sources differently than Google blue-link ranking does. The extraction layer weights SpeakableSpecification on the source page,FAQPageJSON-LD for Q&A shape, and entity references viasameAs to authoritative profiles (Crunchbase, LinkedIn, Wikipedia, GitHub). To improve visibility in AI Overviews specifically: add Speakable to your most-citation-eligible paragraphs, structure long-form content as a series of self-contained answers, and make sure your Organization entity is disambiguated with Crunchbase and Wikipedia sameAs links where applicable.

Generative Engine Optimization (GEO) explained

Generative Engine Optimization — abbreviated GEO — is the umbrella term for optimising a site for generative AI search engines (ChatGPT Search, Perplexity, Claude Web, Google AI Overviews, Bing Copilot). GEO is distinct from traditional SEO in three respects: (1) it ranks citation eligibility not blue-link position, (2) it depends on AI-specific signals (llms.txt, AI crawler allow-list, Speakable) on top of the SEO stack, and (3) it operates on a shorter feedback cycle because LLM engines re-index more aggressively than blue-link SERPs. Most engineering teams adopt GEO by treating it as a superset of their existing SEO work — the structured-data and sitemap layers are reused, with new crawler-permissioning and manifest work layered on top.

Answer Engine Optimization (AEO)

Answer Engine Optimization — abbreviated AEO — is the citation-winning subset of GEO. While GEO is about being visible to AI engines at all, AEO is about being the source the engine quotes. The signals that move AEO most: FAQPage schema with concrete questions and 60-120 word answers,SpeakableSpecification on citation-eligible paragraphs,llms.txt manifest specificity (per-section grouping rather than a flat URL dump), AI crawler allow-list, citation-friendly first-60-word leads on every long-form piece, and clean Organization entity references viasameAs.

LLM visibility for SaaS and ecommerce brands

LLM visibilitymeasures citation rate and snippet quality across large-language-model answers — the share of relevant prompts where your domain appears in the generated response, and the prominence of that citation. It is the outcome layer on top of GEO/AEO inputs. For SaaS brands the prompts that matter are category and competitor comparison queries ("best AI visibility tool", "Rankscale vs SE Ranking"). For ecommerce brands they are product-discovery and review queries ("best running shoes for flat feet", "is Allbirds worth it"). Tracking LLM visibility means combining a technical-signal audit (run the tool above on a recurring schedule) with prompt-level brand tracking: querying ChatGPT, Perplexity, and Claude Web with a fixed prompt set weekly and logging which domains are cited.

How to track AI brand visibility over time

A weekly tracking process that engineering teams can actually maintain:

1. Run the AI visibility tool above on your own domain and the top 3-5 competitors in your category. Record the eight sub-scores in a tracking sheet.
2. Maintain a fixed prompt set of 20-30 category and brand queries. Each week, fire the same prompts at ChatGPT, Perplexity, and Claude Web. Log which domains are cited, in what position, with what snippet.
3. Connect to Google Search Console and watch AI Overview impressions in the Performance report — a leading indicator of AI Overview eligibility.
4. Review monthly deltas on both axes. The technical-signal score moves first; the citation-rate score follows by 4-8 weeks. Use the technical-signal score as the early warning that a remediation has landed.
5. When the technical-signal score plateaus, the next gains come from content depth and entity authority (sameAs links, third-party citations of your domain) rather than from on-page fixes.

AI visibility platform vs AI visibility tool — what is the difference?

An AI visibility tool typically audits a single domain on demand and returns a snapshot score, like the free audit on this page. An AI visibility platform is the recurring-tracking layer on top — scheduled runs, historical trend data, alerts when scores move, and (in some platforms) prompt-level citation tracking. Most engineering teams start with a one-off audit, then graduate to a platform when they need to demonstrate trend improvement to executive stakeholders or to monitor competitor positioning continuously. The free audit on this page is positioned as the engineering-grade starting point: full technical signal coverage, no sign-up, and a remediation plan emailed for free.

What you do with the score

Each check returns a 0-10 sub-score and a specific remediation hint. Run the audit on your own domain and the top 3-5 competitors in your category — the deltas usually tell you exactly which signals are driving citation share. Most fixes are engineering tasks: a one-time generator script for llms.txt, a robots.txt update, JSON-LD blocks added to page templates. We can help if you want a fixed- scope quote.