Site audit basics for AEO

VStok site audit crawls up to 500 pages on your domain and evaluates signals answer engines use to discover, parse, and cite content. Categories include robots.txt rules, XML sitemap presence, llms.txt, structured data (schema), meta tags, content quality markers, and site structure. Each finding has a severity: critical, warning, or ok.

Run site audit from the project sidebar after your visibility audit baseline exists. On Free you get one site audit per month; Starter includes five, Growth twenty. Results appear separately from prompt metrics — a perfect visibility score can coexist with crawl blockers that will hurt future citations. Hybrid projects should run both site and app listing audits. The crawl respects robots rules — blocked pages are reported, not fetched.

Robots.txt is the first gate. VStok flags rules that block OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot, and other AI crawlers listed in your report. A single Disallow on /blog can exclude your best comparison content from model retrieval. Fix by allowing AI bots explicitly or narrowing disallows to truly private paths.

Sitemap issues — missing index, stale lastmod, or orphan strategic pages — reduce discoverability. llms.txt is an emerging convention: a machine-readable file at /llms.txt that points models to canonical docs, policies, and product overviews. VStok checks presence and basic validity. Add llms.txt with links to your most citation-worthy pages; regenerate after major site launches. Keep the file under a few dozen links so it stays maintainable.

Structured data helps models extract entities: Organization, Product, SoftwareApplication, FAQPage, and Article types matter most for B2B SaaS. The audit lists pages missing JSON-LD or with parse errors. Prioritize money pages — pricing, product, integrations, and comparison URLs — before blog archives.

Meta title and description gaps weaken how snippets get summarized when pages are retrieved. Content checks flag thin pages, missing H1 hierarchy, and blocked canonical mismatches. Answer engines prefer clear, factual paragraphs over keyword-stuffed blocks. Pair on-page fixes with FAQ sections that mirror your tracked prompts in natural question-and-answer form. Use FAQPage schema on help center articles for double signal.

The Recommendations tab merges site audit issues with visibility and citation gaps. Items like fix_robots_txt and add_llms_txt link back to crawl evidence. Work critical issues first — they block entire sections of the site from AI crawlers. Warnings are often quick wins: single-page schema additions or meta rewrites.

Re-run site audit after deploying fixes to confirm severity downgrades. Allow a few days before expecting visibility score lifts; models need time to recrawl and refresh citations. Document deploy dates next to audit timestamps so your team correlates technical changes with dashboard movement. Export audit summaries for developers via public reports or PDF on Agency plans. Assign each critical issue a ticket owner in your issue tracker with a target ship date.