Search engine volume itself is now expected to shrink — which means the old audit checklist, built entirely around ranking, is no longer the whole test.
Gartner predicts search engine volume will drop 25% by 2026 due to AI chatbots. That is not a forecast about a niche behavior shift — it is a forecast about where the traffic your business depends on is going next, and it is already underway. A shrinking share of your audience is scrolling ten blue links. A growing share is asking ChatGPT or Perplexity a question and accepting whatever answer comes back, citations and all.
"Gartner predicts search engine volume will drop 25% by 2026 due to AI chatbots."
— Gartner, cited in Hashmeta AI GEO researchTraditional SEO audits were built to answer one question: will this page rank? That question still matters — organic ranking remains the raw material AI engines draw citations from — but it is no longer sufficient on its own. A page can rank respectably and still be structurally invisible to a language model synthesizing an answer, because ranking and citability are measured by different mechanics entirely.
Your goal isn't just visibility anymore — it's citability.
Citability is earned at the level of individual technical and content decisions — decisions most teams have never audited because the checklist to audit them didn't exist yet. We built one. It's organized into five categories of six points each, covering everything from robots.txt to author bylines, and it's exactly what we run against a domain before recommending a single piece of new content. Work through it in order, and be honest about what's actually failing versus what merely looks fine.
Technical Foundation
POINTS 1–6If AI crawlers can't reach or parse your pages cleanly, nothing else on this list matters. Start here.
Structured data. Ship Product, FAQ, HowTo and Article schema markup on every key page so AI engines have structured ground truth to lift, not just prose to interpret.
XML sitemaps. Keep an up-to-date XML sitemap submitted to Google Search Console so every page you want cited is fully discoverable and indexable.
robots.txt. Audit robots.txt to confirm GPTBot, ClaudeBot, PerplexityBot and Google-Extended are explicitly allowed — a single blocked crawler is a silent exit from that engine's citation pool.
Page speed. Get Core Web Vitals into the "good" range sitewide; slow pages get crawled less frequently and cited less often, regardless of content quality.
Mobile responsiveness. Confirm every page renders and reads cleanly on mobile — most crawlers, and a growing share of human verification traffic, evaluate the mobile version first.
HTTPS. Verify HTTPS is enforced sitewide with no mixed-content warnings; unencrypted or broken-certificate pages are filtered out of citation-worthy source pools.
Content Structure
POINTS 7–12AI engines lift content, they don't interpret it — the more machine-parseable your structure, the more extractable your answers.
Heading hierarchy. Use a clean, logical H1–H3 hierarchy with question-based H2s; well-structured sections are cited roughly 2.8× more often than equivalent unstructured prose.
Direct answers. Open each section with a direct, extractable answer in the first sentence or two, before expanding into supporting detail.
Lists. Break stepwise or comparative information into bulleted or numbered lists — the format AI engines most reliably lift verbatim.
Tables. Present comparisons, pricing or spec data in HTML tables rather than paragraphs, so the structure is machine-parseable at a glance.
Paragraph density. Keep paragraphs short — two to four sentences; dense, unbroken blocks of text are harder for models to extract cleanly.
Visual optimization. Add descriptive alt text and captions to charts, diagrams and screenshots so visual content carries the same citable meaning as the surrounding text.
Authority & Trust
POINTS 13–18AI engines preferentially cite claims and sources they can verify. This category is your E-E-A-T footprint, made explicit.
Author credentials. Byline every article with a named author, their credentials, and Author schema markup — E-E-A-T signals AI engines weight at the author level, not just the domain level.
Publication dates. Display and mark up a visible dateModified on every page, keeping cornerstone content refreshed at least once every 12 months.
Source citations. Attribute every stat, benchmark or claim to a named, verifiable source; unattributed assertions get filtered out of citation surfaces.
E-E-A-T signals. Surface Experience, Expertise, Authoritativeness and Trust explicitly — case studies, first-party data and named expert commentary all count.
Organizational transparency. Maintain a clear About page, a verified Wikidata entry and a Google Knowledge Panel so AI engines can resolve exactly who you are.
Quality outbound links. Link out to authoritative, independently verifiable sources rather than only internal pages — it signals the same rigor you're asking AI to trust you for.
Answer Optimization
POINTS 19–24Write for the exact question a buyer — or the model synthesizing their answer — is actually asking.
FAQ sections. Add a genuine FAQ section addressing the real long-tail questions buyers ask, marked up with FAQPage schema.
Question-based headings. Phrase key H2s and H3s as the literal question a buyer would type or ask an AI assistant.
Multiple answer formats. Answer important questions in more than one format — a sentence, a list, and a table — so whichever an engine prefers to lift, it finds one.
Contextual definitions. Define key terms and jargon plainly the first time they appear, rather than assuming the reader — or the model — already knows them.
Numerical data. Back claims with specific numbers, percentages and benchmarks rather than vague qualifiers like "many" or "significant."
Step-by-step instructions. Present processes as explicit numbered steps rather than narrative paragraphs, formatted for HowTo schema where relevant.
Multi-Platform Visibility
POINTS 25–30Your own domain is one source among many an AI engine draws from — this category is about showing up everywhere else it looks too.
Content distribution. Syndicate cornerstone content to relevant industry publications and partner sites, widening the surface area AI engines can draw from.
Social media. Maintain active, verified social profiles that reinforce brand entity signals across the platforms models are trained on.
Multimedia. Publish video and audio content with transcripts, since transcribed multimedia is indexable and citable in the same way as text.
Industry citations. Pursue mentions and backlinks from category-trusted publications and associations — the earned-media signal AI training data weights most heavily.
Knowledge bases. Keep your Wikidata entry, Crunchbase profile and other structured knowledge-base listings accurate and current.
API accessibility. Where relevant, expose structured data via public APIs or feeds — giving AI systems a direct, machine-readable path to your information beyond crawling HTML.
Score yourself against all 30 points.
Most sites pass fewer of these than they expect — usually the technical foundation and authority categories are where the surprises hide.
The Window Is Still Open
None of these 30 points is exotic. Individually, most are a few hours of engineering or editorial work. Collectively, they are the difference between a domain AI engines can confidently cite and one they quietly route around in favor of a competitor who did the work first. Run this checklist honestly, fix what's failing in the order it appears above, and re-run it every quarter — citability isn't a one-time project, it's a standing to maintain.
Want us to run the full 30-point audit against your own domain, scored and prioritized? Get a free GEO audit and we'll show you exactly where you stand.