40-Point Checklist: get cited by ChatGPT
The 40 points we check when we audit a page on a €3,000/month engagement. We're handing them over, block by block, with each one's impact level.
Optimisé pour être cité dans
This checklist lists the 40 points we check during a GEO audit so a page gets cited by ChatGPT, Perplexity or Gemini. Eight blocks, five points each, from the most technical to the most editorial — each with its real impact level. You tick the boxes, you see your gaps, you know where to start.
Why this checklist rather than yet another guide
When we audit a page to find out whether an AI can cite it, we don't go by feel. We work through a list. Always the same one. Because LLM citability isn't a mystery: it's a series of concrete signals, each verifiable one by one.
The problem is that nobody publishes this list. People tell you "GEO matters," they sell you a service, but they never show you the actual checks. Here, we're giving you the whole thing. It's what we verify on a €3,000/month engagement.
Why do it? Because a prospect who understands what we check makes a better client. And because half of these points, you can fix on your own. The other half is where we come in.
The 8 control blocks
Each block has 5 points. The order isn't random: we start from the technical foundation (without it, nothing gets read) and end with measurement (without it, you're flying blind).
1 · Technical
Is the page accessible, fast, indexable and rendered without blocking JavaScript? If an AI bot can't load your content, everything else is pointless.
2 · Content
Do you answer a clear intent, right away, without beating around the bush? AIs cite direct answers, not introductions.
3 · Schema
Does your structured data tell the machine who you are and what the page is about? FAQPage, Article, Organization: the markup LLMs read first.
4 · Off-site
Does your brand exist anywhere beyond your own site? Reddit, YouTube, press, third-party mentions — AIs cross-check sources before citing.
5 · Passage citability
Are your paragraphs extractable as-is? A self-contained passage of 130 to 170 words gets cited. A 600-word wall of text doesn't.
6 · AI accessibility
Do you explicitly allow AI bots to read your site? robots.txt, llms.txt, headers — many block AIs without realizing it.
7 · Author authority
Can we tell who wrote it, and why to trust them? E-E-A-T applied to people: bio, signals, consistency across the web.
8 · Measurement
Can you prove you're being cited, and track the trend? Without tracking, you'll never know if your efforts are paying off.
The 40 points in detail
Block 1 — Technical
The page returns a 200 status, isn't accidentally noindexed, and appears in your sitemap. Check it for real, not from memory.
The main content is present in the raw HTML, not injected client-side. Most AI bots don't render heavy JS.
The page loads fast, on mobile as on desktop. A crawler that times out won't read you.
A clean URL, free of junk parameters, that doesn't change. AI citations point to durable URLs.
Everything runs over HTTPS, with no redirect chains that lose the bot along the way.
Block 2 — Content
The page answers the main question within the first 80 words, before any context. That's what the AI extracts.
One page = one question = one answer. Not three mixed topics the machine can't isolate.
Your key terms are defined on the spot, "X is…" style, without sending readers elsewhere. A format LLMs love to reuse.
You cite precise figures, sourced and dated. An AI would rather cite verifiable data than a generality.
A real update date, and content that's genuinely been refreshed. AIs weigh freshness.
Block 3 — Schema
Your entity is marked up: who you are, your logo, your official profiles. The foundation of machine trust.
Type, author, date, headline declared in JSON-LD. The machine knows what it's reading.
Your questions and answers are cleanly marked up. A format an AI can reuse directly in a response.
Structured breadcrumbs help the machine place the page within your site hierarchy.
The markup passes validation without blocking warnings. Broken schema is useless, even counterproductive.
Block 4 — Off-site
Your name appears on sites you don't control. AIs cross-check before citing.
Your topic is discussed where AIs draw community signal. No spam — real presence.
A video covering your topic widens your citable surface. Generative engines pull from video.
Name, address, contact identical everywhere. An inconsistency casts doubt on your identity.
Credible sites point to you. Link building remains an authority signal AIs inherit from Google.
Block 5 — Passage citability
Your key blocks are calibrated to be extracted as-is. The format LLMs reuse most readily.
Each passage makes sense out of context, with no "as we saw above." The AI extracts in chunks.
Your H2/H3 mirror users' real questions. The machine matches question to answer.
Comparisons and steps are in lists or tables. Formats that are easy to reuse in a response.
Zero hollow paragraphs between the question and the answer. Padding dilutes the citable passage.
Block 6 — AI accessibility
GPTBot, PerplexityBot, Google-Extended and the rest aren't blocked by mistake. Pitfall number one.
An llms.txt that guides AIs toward your key content. First mover: few French sites have one.
The citable content is readable without a login or cookie wall that blocks the bot.
A markdown or text version of your key pages makes extraction easier for AIs.
Your firewall / CDN doesn't return a 403 to legitimate AI bots. Check it in your logs.
Block 7 — Author authority
Every piece carries a real byline, not "the editorial team." AIs favor attributable sources.
The author bio establishes legitimacy through facts, not adjectives.
The author is marked up as Person, linked to their profiles. The machine cross-references the identity.
Same name, same expertise, same topics elsewhere. Consistency builds authority in an AI's eyes.
The content shows lived experience, real cases, original data. The "E" in E-E-A-T that AIs reward.
Block 8 — Measurement
You regularly test whether ChatGPT, Perplexity and Gemini cite you on your target queries.
You identify, in GA4, the traffic coming from generative engines. It exists, and it can be measured.
You know who's being cited in your place, and why. The target is defined by comparison.
You track your 10 to 20 business queries, not a fuzzy average. Granularity guides action.
You revisit these points every month and fix what's off. GEO is steered, not set once.
Who this checklist is for
The marketing or SEO lead who notices their brand doesn't show up in AI answers and wants to understand, concretely, why. The B2B SaaS or SME founder who wants to assess the state of their pages before committing a budget. Anyone who's tired of "you need to do GEO" without being told what to check.
These 8 blocks map to the four stages of the LUWIZ Method: Diagnosis (technical and accessibility blocks), Foundation (content and citable passages), Authority (off-site and author), Steering (schema and measurement). The checklist is our diagnosis placed in your hands.
Get the full checklist
You receive the printable PDF version and the interactive Notion version — the one where you tick off your 40 points and track your progress. Email only, we ask for nothing else.
Once you have the checklist in hand, two paths forward. If you want the tools to execute, the GEO audit template pack gives you the ready-to-use robots.txt and the schema library. If you'd rather have us run the 40 points on your site directly, request your free GEO audit — report in 48h. And to understand the full service behind it, see our GEO agency page.
We run the full checklist on your site and deliver the prioritized report in 48h. Three slots a month.
Questions fréquentes
To verify, point by point, whether a page is technically and editorially ready to be cited by ChatGPT, Perplexity or Gemini. It's the same protocol we run in a paid GEO audit, condensed into 40 checks across 8 blocks, each with an impact level. You tick the boxes, you see your gaps, you prioritize.
No. Half the points are content and editorial — rewriting passages, heading structure, author signals. The other half is technical (schema, robots.txt, AI bot accessibility) but every point is stated in plain language, with what to check and why it matters to an AI.
It gives you 80% of the diagnosis on your own. What it doesn't do: prioritize for your specific sector, compare your page against your real competitors, and estimate the quantified potential. That's what we add in a personalized audit. The checklist is the honest starting point.
Yes. We revise the 40 points continuously as we run audits and as the generative engines evolve. The version you receive is the one we're applying on engagements this week, not a frozen PDF from two years ago.
Envie d'aller plus loin que la ressource ?
Un audit GEO personnalisé de votre site, gratuit, livré en 48h.
Réponse sous 24h · Sans engagement · contact@luwiz.io