Last updated July 2026.
The short answer
Run this checklist before any GEO programme. Many brands that report zero AI visibility have a technical blocker on item 1. Fix the access issues first, then the format issues. Citation footprint and third-party presence take longer but move the biggest needle. Studies consistently find that the large majority of AI citations come from third-party sources rather than brand-owned websites, so a programme that focuses only on your own pages will plateau early.
How to use this checklist
Each item is a yes/no verification with the rationale behind it. Work through the six sections in order. A “no” answer is a gap. Prioritise Section 1 (crawler access) and Section 3 (content format) first. They are the fastest to fix and often the source of the biggest gaps.
The summary table at the end maps each section to the metric it moves.
Section 1: AI Crawler Access (items 1 to 5)
Your content cannot be cited if AI crawlers cannot reach it. According to OtterlyAI’s 2026 AI Citation Economy report, 73% of sites have technical barriers that prevent AI crawler access, including robots.txt blocks, CDN security rules that reject non-browser user agents, and JavaScript rendering issues.
Item 1: Does your robots.txt file allow the major AI crawlers?
Check for disallow rules targeting GPTBot (OpenAI), Google-Extended (Google), ClaudeBot (Anthropic), PerplexityBot, and cohere-ai. A blanket Disallow: / for any of these blocks that engine from indexing your content. Robots.txt is the single most common technical cause of AI invisibility.
Item 2: Does your CDN or WAF allow AI crawler user agents through?
Some CDN security configurations and bot-management rules reject requests that do not come from a standard browser. Check your WAF ruleset and rate-limiting configuration for user-agent pattern blocks that would catch crawlers. Your robots.txt can be open while the CDN silently drops every request.
Item 3: Do your key pages render without JavaScript?
Most AI crawlers do not execute JavaScript. A page that loads in a browser but delivers an empty DOM to a crawler is effectively invisible. Test your most important pages by disabling JavaScript in a browser or using a tool that fetches raw HTML. If the main content is absent, you have a rendering barrier.
Item 4: Are your most important pages included in your sitemap and indexed by Google?
AI engines draw heavily from pages they can discover through standard crawl paths. If a page is not indexed by Google, it is unlikely to be indexed by AI crawlers that follow similar discovery signals. Confirm sitemap coverage and Google indexing status for your key category, product, and editorial pages.
Item 5: Is your server response time under two seconds for raw HTML?
Slow server responses cause crawlers to time out and move on. This matters more for AI crawlers than for browsers because they rarely retry. Confirm your pages return a response quickly for non-browser user agents, not just cached browser responses.
Section 2: Schema Coverage (items 6 to 10)
Schema is not the most powerful lever in GEO, but it is a verifiable yes/no signal. An Ahrefs study tracking 1,885 pages that added JSON-LD schema found no statistically significant uplift in AI citations on its own. Schema is necessary, not sufficient: the underlying prose still has to be retrieval-friendly, and the page must be crawlable.
Item 6: Does each key page have at least one relevant schema type implemented?
Article, FAQPage, HowTo, and Product schemas are the types most consistently associated with citation presence. At minimum, your editorial pages should carry Article schema with an accurate dateModified field. Product pages should carry Product schema with price, availability, and review aggregate where applicable.
Item 7: Does your FAQPage schema contain the same direct-answer prose as the visible page?
Schema that diverges from visible content is flagged by quality systems. The question-and-answer pairs in your FAQPage schema should match what a reader actually sees. Do not use schema to stuff answers that do not appear in the body.
Item 8: Is your schema valid with no critical errors in a validator?
Invalid schema is ignored. Run your key pages through Google’s Rich Results Test or Schema.org validator. Fix any critical errors before worrying about schema type coverage.
Item 9: Does your Author schema include byline, credentials, and a link to an author bio page?
E-E-A-T signals (Experience, Expertise, Authoritativeness, Trustworthiness) influence which sources AI engines trust. Author schema that includes a named person, their credentials, and a consistent bio page strengthens the signal that the content comes from a real expert.
Item 10: Do your most-cited pages carry datePublished and dateModified in schema?
Freshness is a signal. Pages without date metadata are harder for AI systems to assess for recency. Confirm that datePublished and dateModified are accurate, not static or missing. An incorrect dateModified (one that never changes) is worse than no date at all, because it signals stale content.
Section 3: Content Format (items 11 to 16)
Format determines whether an AI engine can extract a clean, citable answer. According to Kevin Indig’s 2026 analysis of 18,012 verified ChatGPT citations (reported by Search Engine Land), 44.2% of citations were drawn from the first 30% of a page’s content. Pages that bury the conclusion get paraphrased out; pages that front-load a self-contained direct answer get reproduced and credited.
Item 11: Does every target page open with a direct-answer paragraph in the first 200 words?
The retrieval layer grabs the first clean passage that looks like an answer. Write the answer first, then the context. A page that opens with “In this article we will explore…” is competing against a page that opens with a one-sentence answer. The second page wins the citation.
Item 12: Are your headings in a strict sequential hierarchy (H1, then H2, then H3)?
According to AirOps’ July 2025 analysis of over 12,000 ChatGPT-cited URLs, 68.7% of pages cited by ChatGPT used a sequential heading structure, compared to only 23.9% of Google’s top-ranked pages for the same queries. A flat heading structure (H2, H2, H2 without H1) is a format signal that depresses citation rates.
Item 13: Do your key pages include at least one structured list where the content calls for it?
According to AirOps’ analysis, nearly four out of five pages cited by ChatGPT include at least one structured list, compared to just 29% of Google’s top-ranked pages. Lists let the extraction layer pull a clean enumeration without paraphrasing. This applies to steps, criteria, features, and options but not to everything. Do not add lists where prose is the right format.
Item 14: Do comparison or multi-item pages include at least one Markdown or HTML table?
According to AirOps Research (April 2026), comparison pages containing three or more HTML tables earn 25.7% more AI citations than those without, specifically for head-to-head product comparison queries. If you have a comparison page, a tools roundup, or a feature matrix, a table is not optional.
Item 15: Is the reading level of your key pages at or below 10th grade?
AI engines reproduce content that is easy to parse and re-use in a generated answer. Dense, jargon-heavy prose is harder to extract cleanly. Aim for short sentences, short paragraphs, and plain language. Tools like Hemingway App give you a reading-level score in seconds.
Item 16: Do your pages avoid thin introductions that delay the substantive answer?
A 200-word introduction that recaps why the topic matters before getting to the answer pushes the citable content past the 30% mark where most citations cluster. Cut preamble. Get to the answer fast.
Section 4: Citation Footprint (items 17 to 20)
Citation footprint is the set of source pages that AI engines pull from when answering queries in your category. Studies consistently find that the large majority of AI citations come from third-party sources rather than brand-owned websites, with figures ranging from roughly 77% to over 85% depending on methodology and the AI platforms studied. You cannot fix your citation footprint only by optimising your own site.
Item 17: Do you know which source pages AI engines cite for your top-priority category queries?
Run your five most important category queries through ChatGPT, Perplexity, and Google AI Overviews. Record every source URL cited. This is your citation map. If none of those URLs are your domain, you have a footprint gap, not just an optimisation gap.
Item 18: Is your brand mentioned on at least three of the top-10 domains cited for your category queries?
Generative engines lean on a concentrated set of authoritative sources. If the top-cited domains for your category (review sites, industry publications, analyst reports) do not mention your brand, your own content cannot compensate. Identify those domains and build a presence on them.
Item 19: Do you have at least one mention or review on G2, Capterra, or a category-specific review platform?
Review platforms are consistently over-indexed in AI citations for commercial queries. A brand with no review presence on the platforms AI engines trust for your category is structurally disadvantaged. One quality review page on a trusted platform can move citation share faster than many on-page fixes.
Item 20: Have you corrected any factual inaccuracies AI engines state about your brand?
If AI engines are citing incorrect information about your product, pricing, or positioning, that misinformation feeds back into future answers. Identify the source of the inaccuracy (often a misquoted article or outdated third-party page) and correct it at the source. Structured correction workflows are part of a complete GEO programme.
Section 5: Entity and Knowledge Graph Signals (items 21 to 23)
AI engines build representations of brands, concepts, and relationships from the structured and unstructured data they index. Strong entity signals make your brand recognisable across contexts.
Item 21: Is your brand listed on Wikipedia or Wikidata with accurate, up-to-date information?
Wikipedia is consistently one of the most-cited sources in AI-generated answers. If your brand is notable enough for a Wikipedia entry, that entry should be accurate and maintained. Wikidata entries provide machine-readable entity data that feeds directly into knowledge graph representations.
Item 22: Is your brand’s Google Business Profile (or equivalent) complete and accurate?
Local and regional AI answers draw on structured business data. A complete, accurate, and regularly maintained business profile strengthens the entity signal for your brand across engines that source from Google’s knowledge graph.
Item 23: Is your brand name used consistently across your own site, press coverage, and third-party listings?
Inconsistent brand name formatting (abbreviations, rebrands, informal names) fragments the entity signal. AI engines associate citations across sources by brand name. A fragmented name means fragmented credit. Audit your name across your own pages, press mentions, partner sites, and listing directories.
Section 6: Freshness and Third-Party Presence (items 24 to 27)
Freshness signals tell AI engines that content is current and worth citing. Third-party presence determines whether sources the engines already trust carry your brand. These are the two areas where most brands have the largest long-term gap.
Item 24: Have your key category pages been updated within the last 90 days?
AI engines favour recent content, particularly for fast-moving categories. A page that has not been touched in 18 months competes poorly against one that was refreshed last month. Update the dateModified, add any new data or context, and re-submit to Google Search Console to trigger re-crawl. Do not just change a date without substantive content changes.
Item 25: Is your brand mentioned in earned media published in the last six months?
Earned media (articles, reviews, mentions in publications you did not pay for) is the primary channel through which brands build the third-party citation footprint that AI engines weight most heavily. If your last earned media mention is older than six months, you have a freshness gap in your third-party presence. Prioritise placements on the domains your citation map (item 17) identified as over-cited.
Item 26: Do you have at least one published case study, data report, or original-research piece that third parties can cite?
Original data is one of the most citable content types. Proprietary benchmarks, survey results, and case studies give journalists, analysts, and bloggers something worth linking to and citing. If your site has no original data, your third-party citation opportunities are limited to product reviews and brand mentions.
Item 27: Do you track your citation share across at least three AI engines on a weekly or bi-weekly basis?
You cannot improve what you do not measure. Citation share is the metric that tells you whether your GEO programme is working. Tracking it across multiple engines matters because, according to Profound’s analysis of 100,000 prompts, only about 11% of domains are cited by both ChatGPT and Perplexity. A brand that appears in ChatGPT answers may be nearly absent from Perplexity, and vice versa.
Section summary: what each area moves
| Audit section | Primary metric it moves | Fastest to fix? |
|---|---|---|
| 1. AI Crawler Access | Indexation by AI engines | Yes: hours to days |
| 2. Schema Coverage | Structured-data retrieval signal | Yes: one to two sprints |
| 3. Content Format | Citation rate from indexed pages | Moderate: requires rewriting |
| 4. Citation Footprint | Share of model in category queries | Slow: requires outreach |
| 5. Entity Signals | Brand recognition across contexts | Moderate: one-time setup |
| 6. Freshness and Third-Party Presence | Long-term citation share trajectory | Slow: ongoing programme |
Tools that cover these audit areas
No single tool covers every layer of a GEO audit, but several address specific sections well.
Crawler access and on-page audit: Otterly.AI includes a GEO Audit Engine that checks any URL against 20-plus on-page citation-readiness factors, covering robots.txt signals, rendering issues, and schema gaps. Scrunch goes further at the infrastructure level with its Agent Experience Platform, which serves AI-optimised content directly to LLM crawlers.
Citation footprint mapping: Profound provides visual citation source maps that show exactly which URLs AI engines pull from for your category queries. That is the input you need for items 17 and 18 above. Ahrefs Brand Radar gives competitive citation data across six AI platforms drawn from a database of more than 400 million prompts, which is useful for benchmarking your footprint against competitors.
Citation share tracking: Knowatoa and Peec AI both track brand mentions across AI engines and give you the weekly citation share data item 27 requires. Temso covers monitoring across eight engines alongside gap diagnosis and content execution from a single subscription, which is useful if you want to run the full programme in one tool rather than assembling separate audit, tracking, and execution platforms.
See the full ranking of GEO tools at /rankings/geo-tools or the /glossary for definitions of citation share, share of model, and entity signals.
Run the audit, then prioritise
Work through all 27 items. Mark each one pass or fail. Then prioritise in this order:
- Fix any “no” in Section 1 (crawler access) before anything else. A blocked crawler makes every other improvement invisible.
- Fix any “no” in Section 3 (content format) next. Format gaps are fast to fix and have a direct effect on citation rate from pages that are already indexed.
- Address Section 2 (schema) and Section 5 (entity signals) in parallel. These are largely one-time setup tasks.
- Build a plan for Section 4 (citation footprint) and Section 6 (freshness and third-party presence). These require ongoing effort and produce results over weeks and months, not days.
The brands that declare themselves invisible in AI search without running this audit usually find a Section 1 or Section 3 problem that explains most of the gap. Start there.