SEO & AI Search

AI Visibility

A growing share of buying research never touches a results page: people ask ChatGPT, Perplexity, or Google's AI Overview and read a synthesized answer with a handful of citations. Either your site is one of those citations or you don't exist in that conversation. Optimizing for this is called GEO — generative engine optimization — and it is mostly not the same work as classic SEO.

The good news: because almost nobody does this work yet, it's the rare channel where small sites can outrun incumbents by simply being well-prepared.

Running a check

Open SEO → AI Visibility and hit Run check. The audit probes your site the way an AI crawler would — fetching root files, parsing page HTML, testing extraction — and returns an overall score, metric tiles (pages checked, criticals, warnings), and an AI-written summary.

A five-point summary shows where you stand at a glance: llms.txt file · AI crawler access · Schema.org structured data · Content extractability · Citation signals, each marked passed or needs attention.

The five finding categories

1. Crawler config — are AI bots allowed in? Many sites block GPTBot and friends by accident, often via a CDN default. The panel includes a primer on the crawler types worth distinguishing — training crawlers, search/retrieval crawlers, user-triggered fetchers — and a recommended robots.txt policy, since blocking training bots while admitting search bots is a legitimate choice you should make deliberately.
2. AI index files — llms.txt and llms-full.txt, the emerging convention for handing LLMs a clean map (and full text) of your site.
3. Discoverability — sitemap coverage, clean descriptive URLs versus opaque IDs, canonical hygiene.
4. Content form — can an LLM extract your content? Text rendered only by JavaScript, thin pages, and missing Article/FAQ schema all reduce what an engine can quote.
5. Anti-patterns — things that actively hurt: cloaking (serving crawlers different content), FAQ-schema stuffing, boilerplate-heavy HTML.

Every finding explains why it matters and how to fix it. There's also a full check history to track your score over time.

Fixing

Most findings carry a one-click fix: add llms.txt, add llms-full.txt, edit robots.txt, add a sitemap, clean URLs, remove over-used FAQ schema, and more. Fix all with AI hands the whole list to the agent — same mechanics as the Site Audit fixer, including the bound project folder and reviewable diffs.

The visibility score

Audits tell you whether you're quotable. The visibility scan tells you whether you're being quoted: it measures real LLM answer data for your domain (as a citation source) and your brand (as a mention), merged into a 0–100 visibility score with an executive summary.

Note: ChatGPT answer data currently covers the US/English market only; other markets measure Google AI Overview. Scans are deduplicated and budget-capped, so re-running frequently doesn't burn anything.

What actually moves the needle

In rough order of return on effort:

1. Unblock AI search crawlers (five minutes, often the whole problem).
2. Ship llms.txt and llms-full.txt — one-click here.
3. Make your best pages extraction-friendly: real HTML text, clear headings, Article schema.
4. Publish content structured as answers; the Article Studio's citation-backed articles are exactly the shape AI engines like to cite.

All checks here are free, like every diagnostic. Run it monthly; this field shifts fast.