GEO Strategy: On-Page GEO

As the digital landscape shifts to include AI-driven search, a new discipline has emerged: Generative Engine Optimisation (GEO).

While traditional SEO focuses on keywords, links and Googles EEAT principle, GEO focuses on becoming the "source of truth" for AI models like Gemini, Perplexity, and OpenAI’s ChatGPT.

To stay visible, your on-page strategy must consider existing SEO principles whilst optimising for comprehensiveness, authority, and citability, as this is essentially what the LLM will show in the ‘chat’.

‘AI models prefer content that reads like a complete, trustworthy answer to a specific question'.

To optimise for prompts:

- Write clear, direct answers to likely questions

- Add FAQ sections with short factual responses

- Use subheadings that match how user would phrase a prompt

(Source: SEMRush, 2025)

AI engines are designed to scan and summarise. If your page is a "wall of text," an AI might miss the nuances of your argument.

On-page GEO requires a modular structure. Alongside on-page SEO optimisation, you can pay particular attention to the following GEO on-page elements that make it as easy as possible for LLM’s to cite information.

Web Page Structure

Begin with a question: Start your post with a question as opposed to a statement, to align with human’s asking LLM’s a question. (H1 Tag).

The "Lead-In" Summary: Start your articles with a concise, 2-3 line sentence summary of the primary answer to the question. This provides the AI with a ready-made "snippet" to pull into its response.

Semantic Header Hierarchy: Remember to Use H2 and H3 tags not just for aesthetics, but to create a logical map of the topic. AI models use these headers to understand the relationship between different sub-topics.

Bulletins and Tables: Generative engines love structured data. If you are comparing products or listing steps, use a table or a numbered list. This increases the likelihood of your data being featured in a comparative AI response.

Numbers and Ranges: Generative engines like numbers and ranges, definitive measurements and illustrations of scale to give clear and concise answers to questions. By giving factual information you increase the chance of being cited in results.

Page Formatting

We use the page format to make it as easy as possible for the AI engine to cite the data. From communicating the most important information towards the top of the page, to including small facts in FAQ formatting, infographics and diagrams ‘tell 100 words’.

Schema markup assists AI engines to crawl pages quickly and effectively within their crawl budgets.

The "Inverse Pyramid" Style: Place the most critical information at the top. Since AI models have "context windows" (limits on how much text they process at once), the most vital points must be prominent.

FAQ Blocks: Adding FAQ sections to key pages mirrors the Q&A structure AI engines favour.

Multi Format Support: Charts, infographics and video transcripts provide additional clarity, though always paired with strong on-page text.

Schema Markup: Our primary research document on how to gain brand visibility from ai search showed that 90% of content cited by ChatGPT & Gemini had schema markup present on the web pages.

Language: Direct, Objective, and Technical

AI models are trained to avoid bias and "fluff." To optimise for these engines, your on-page tone should shift toward authoritative neutrality. The goal is to provide the maximum amount of value in the fewest number of words; written like a human. If a paragraph doesn’t add a new fact or a new perspective, it may actually dilute your "GEO score."

Clarity over Creativity: be concise with your content and don’t add repetition or unnecessary ‘fluff’. AI engines are looking for clear and concise information within their limited crawl budgets.

Comparisons & Definitions: AI Engines like clear definitions (‘X is…’) and structured comparisons (‘X vs Y’).

Language: Don't shy away from technical terms. Using the specific language of your industry helps the AI categorise your content as a "high-knowledge" source.

Adjectives like "revolutionary," "best," or "amazing" are often filtered out by AI as subjective noise; which they are. Stick to descriptive, factual language.

Boosting Authority through "Citations and Statistics"

In traditional SEO, content and links were the primary currency of authority. In GEO, alongside on-page GEO signals (this document), positive affirmations on trusted third party websites (Blogs, Forums, Digital Magazines etc) pull the levers.

Recent studies on GEO performance suggest that content featuring specific statistics, expert quotes, and citations is significantly more likely to be included in generative responses.

References and Citations: Adding references and citations shows the LLM that you are a trusted source. Even if you are the expert, linking to peer-reviewed studies or reputable news outlets within your copy signals to the AI that your information is grounded in fact.

Unique Data: If you have original research, highlight it. AI engines prioritise "novel information" that isn't just a rewrite of existing web content.

Designing & writing for a dual audience: AI & Humans

In implementing these strategies, it is important to remain mindful of the dual audience: the AI and the human.

While we optimise for the engine's ability to parse data, we must maintain a narrative flow that respects the reader's intelligence. An overly robotic page might rank well in an AI summary, but it will fail to convert a human user who clicks through to the source, much the same as making a page 'SEO Ugly'.

As time goes by, the brands that thrive will be those that write for humans, but are conscious of AI search signals; backed up by third party approval from trusted sites relevant to their industry.

Written by James Harbottle, JNH Media — GEO and SEO Consultant based in South West England..

Disclaimer: The views and opinions expressed in this article are solely those of the author and do not necessarily reflect the official policy or position of any other agency, organisation, employer, or company. All references and data sources have been cited appropriately. The information provided is for educational purposes only; the author accepts no liability for any results or actions taken based on the findings presented in this post.