In order of impact: first, ensure robots.txt explicitly allows GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, and Google-Extended. Many sites that implemented broad bot blocks during the scraping concerns of 2023 and 2024 are inadvertently blocking AI crawlers. Second, eliminate JavaScript rendering barriers on content that matters for citation. AI crawlers have inconsistent JavaScript execution capability. Server-side rendered content (Next.js App Router, for example) is reliably accessible. Third, fix broken internal links that prevent crawlers from reaching important pages. Fourth, resolve Core Web Vitals failures on top pages, as page quality signals influence AI model training data selection.

Metadata alone is insufficient. Meta descriptions and title tags inform traditional search but AI models primarily extract from body content: the actual sentences on the page. The most impactful changes are in the opening sentences of each section (lead with the answer), the H2 and H3 structure (use question formats), and the body paragraphs (short, direct, citation-ready). Schema markup in the head confirms what the page is about but the content of the page is what gets extracted and cited.

Run four checks. First, search your brand name in ChatGPT and Perplexity and note whether the information returned is accurate and current. Inaccurate AI responses often indicate outdated or conflicting entity signals. Second, check your robots.txt file for blocks on AI crawlers (GPTBot, PerplexityBot, ClaudeBot). Third, validate your schema markup on your five most important pages using Google's Rich Results Test. Fourth, run a Core Web Vitals report in Search Console and note any pages failing on LCP or CLS. Citable's Technical SEO Audit covers all four systematically and delivers a prioritized fix list within 5 business days.

The highest-impact schema types in order: Organization (establishes your brand identity across the knowledge graph), FAQPage (directly maps your content to question-format prompts), Service or Product (establishes what you offer and at what price), Article (for blog and journal content, enables author attribution), and BreadcrumbList (improves content hierarchy understanding). Person schema on About pages strengthens E-E-A-T signals. All schema should be implemented as JSON-LD in the document head, not as microdata inline in content.

Indirectly yes. Core Web Vitals performance is a signal in Google's systems, which influence AI Overviews. For standalone LLMs like ChatGPT and Perplexity, page speed affects how reliably AI crawlers can access and index content during their crawl cycles. Pages that time out or load slowly during crawler visits produce incomplete or missing index entries. Lighthouse 95+ performance is the target for every page Citable builds or optimizes.

llms.txt is an emerging convention (analogous to robots.txt) that provides AI language models with a structured, human-readable summary of your site's most important content and how it should be interpreted. It is not yet universally adopted by AI systems but signals awareness and intent to AI crawlers and the teams building them. Citable deploys llms.txt on every site it builds as a forward-looking signal. For existing sites, adding llms.txt is a low-effort, high-upside technical addition that takes under an hour to implement correctly.