A GEO checklist for AI answers

How do Google’s AI features use your site, and does it need separate optimisation?

GEO is the content and technical work that raises the chance of a page being chosen as a source in generative AI answers. The term spread with the 2023 paper “GEO: Generative Engine Optimization”.

Google’s own guidance is short: there are no additional requirements to appear in AI Overviews or AI Mode. A page needs to be indexed and eligible to be shown in Search with a snippet. No new files, special markup or separate schema are needed. Google adds that meeting the requirements does not guarantee crawling or serving.

That does not make GEO pointless. ChatGPT, Perplexity, Gemini and Claude select short passages from a few sources when they build an answer. The passage that gets picked is the one that answers the question clearly on its own. So the work shifts from making a page findable to making it quotable.

	SEO	GEO
Goal	High ranking and clicks in the results list	Being cited as a source in an AI answer
Success signal	Rankings, click-through rate, sessions	Citations, brand mentions, referrals from AI tools
Content format	A complete page covering a keyword	Answer first, definition sentence, tables, sourced claims
Technical needs	Crawlability, speed, internal links, structured data	The same, plus text visible without JavaScript
Measurement	Search Console, analytics	Search Console plus recurring question tests and referral sources

How do you write a page that gets quoted?

A large language model (LLM) does not read your page from top to bottom when it answers. With a RAG-style setup it splits the page into chunks and uses the chunk that best fits the question. Each section therefore has to make sense on its own.

The checklist:

• Answer first. The first paragraph answers the main question in 40 to 70 words. The backstory comes later.
• Definition sentence. The first sentence under the first heading follows the pattern “X is a Z that does Y”.
• Question headings. Subheadings are the questions readers actually ask. The first sentence under each one answers it.
• Tables. Give comparisons and decision criteria as tables, not paragraphs. Tables keep their meaning when a page is chunked.
• Sourced claims. Every number carries an organisation, a year and a link. The GEO paper’s authors found that adding sources, quotations and statistics improved visibility; the “up to 40%” in the abstract was measured on their own benchmark and varied by domain.
• Freshness. Show an updated date on the page, and actually update it.
• Author and organisation. Make it clear who wrote the page and which organisation stands behind it. Google’s helpful content guidance also stresses the “who, how and why” of content.

What should you check on the technical side?

However good the content, it cannot be quoted if the crawler cannot see it.

• Text without JavaScript. Many AI crawlers read a page without running scripts. Body text should arrive from the server as HTML.
• Crawl permissions. Choose deliberately which bots robots.txt allows. For Google, Googlebot governs Search, while Google-Extended limits use for training and grounding in some other Google systems. OpenAI, Anthropic and Perplexity have their own bot names.
• Snippet controls. nosnippet, data-nosnippet and max-snippet limit how much text can be shown from a page. A stray nosnippet can keep a page out of AI answers.
• Structured data. Markup such as Organization, Article, Product and BreadcrumbList tells machines what a page is. Google’s rule is clear: the markup must match the visible content. FAQ markup is useful for structured content, but do not expect it to produce rich results.
• Canonicals and language. On multilingual sites keep hreflang and canonical tags consistent, or the wrong language version gets quoted.

Where do proposals such as llms.txt stand?

llms.txt is a proposed file at the site root that offers AI tools a Markdown summary of a site and links to its key pages. llmstxt.org describes it plainly as a proposal open to community input. It is not a formal standard.

Google says no special AI text files are needed for its AI features, and other providers have not publicly documented how they use the file. Our advice: adding one is cheap and does no harm, and we publish one on our own site. But do not build your visibility strategy on it. Finish the content and technical items above first.

What changes in practice on WordPress and on a static site such as Astro?

The principles are the same; the implementation depends on the platform.

Item	On WordPress	On a static site such as Astro
Text without JavaScript	Classic themes render on the server; check page builder output	HTML by default; keep client-side components limited
Structured data	SEO plugins such as Yoast or Rank Math, or custom JSON-LD in the theme	JSON-LD in templates, generated from the content schema
Robots and bot access	SEO plugin or a physical robots.txt	robots.txt in the source code
Updated date	The post’s modified date, which the theme must display	An updated field in the content file
Speed	Caching plugin, image optimisation, removing unused plugins	Images optimised at build time, little JavaScript
FAQ sections	Written by hand or with a helper tool	From a faq field in the content schema

Our own site is built with Astro. Body text is visible without JavaScript, and each page’s structured data is generated from its content. On the WordPress side, two of our tools in development focus on parts of this checklist: ContentCheckR reviews content for quality, readability, SEO, tone and compliance risk before you publish, and FAQPilot scans WordPress content and drafts structured FAQ sections with valid FAQPage markup.

How do you track GEO impact?

GEO has no single ranking number. You need to watch several kinds of measurement together:

• Search Console. Traffic from AI Overviews and AI Mode is included in the Performance report under the overall “Web” search type. There is no separate report.
• Referral sources. In analytics, group referrers such as chatgpt.com, perplexity.ai and gemini.google.com into their own channel.
• Question tests. Write a list of 20 to 30 questions your customers would ask. Once a month, put them to several AI tools and record whether you are cited and described correctly.
• Brand mentions. Look at how answers describe your brand, even without a link. A wrong description points to a page that needs fixing.

Where should you start?

Pick the five pages that bring you the most traffic or enquiries and go through them with this checklist. On most sites the first finding is about writing, not technology: the answer is buried in the middle of the page.

At Corrplus we start GEO work with a content and technical audit, then rewrite the priority pages together with you and set up measurement. See our customer experience and digital marketing and WordPress and web platform development services. If you want to scale content production, look at content automation. The checklist matters most for ecommerce sites with many product and category questions, tourism and hospitality sites where visitors compare options, and education sites that answer programme questions every day.