How to write content optimised for AI search

Reading time: 5 min

The way people search online is undergoing a seismic shift. Tools like ChatGPT, Claude, and Gemini are rapidly replacing traditional search engines for millions of users—meaning if your content isn’t optimised for AI, it risks disappearing altogether.

In a recent Evercore analyst poll, 8 percent of respondents said they now use ChatGPT as their main search engine (versus just 1 percent in June 2024), indicating a modest but growing shift away from Google/Bing for core information lookups.

Additionally, Gartner predicts search engine volume will drop 25% by 2026. As of March 2025, a recent survey found that 52 percent of U.S. adults say they use at least one large language model ChatGPT, Gemini, Claude, Copilot, etc. for various tasks, including creating content.

With these stats in mind, how should brands and businesses using content marketing as part of their marketing mix create content that surfaces in LLMs? What are the LLMs looking for? How is it different from how traditional search engines surface content?

To understand this new digital landscape, we spoke with Tom Salvat, co-founder of Elelem.ai and a content intelligence expert to glean practical advice on crafting content that remains visible – and valuable – in an LLM-driven world. Here’s what we found out…

Think questions, not keywords

There will be a shift from keyword searches to question-based searches:

“With traditional search, users typically input keywords to find information. The emergence of LLMs means people now more naturally pose questions. Businesses must create content explicitly structured around answering complete questions rather than targeting fragmented keywords,” says Tom.

Practical Tip: Identify the most common and relevant questions your audience asks, then structure your content around clear, direct answers.

Prioritise clarity and relevance

When it comes to how LLMs rank content, the creator matters less than the content itself. According to Tom:

“It’s less about who created it and more about how effectively the content answers the query.”

Practical Tip: Focus on clarity, depth, and direct relevance when answering your audience’s queries. Avoid fluff and ensure your content directly addresses user intent.

Use content intelligence tools (if budget allows)

To stay ahead, utilise tools designed specifically to help brands adapt to AI-driven search. As Tom advises:

“Content intelligence tools help identify questions customers are asking and how LLMs reformulate these questions to improve retrieval, significantly informing your content strategy.”

Practical Tip: Explore content intelligence tools like Elelem.ai or similar platforms that provide insights into LLM search behaviours and optimise your content accordingly.

Maintain authenticity and originality

In the age of AI-generated content, genuine, human-written content stands out and builds trust. Tom reinforces this:

“High-quality, human-written content should become even more valuable because it builds trust. While AI-generated content can be commoditised, genuinely insightful and original pieces will cut through the noise more effectively.”

Practical Tip: Invest in original, insightful content created by subject-matter experts or journalists to distinguish your brand from AI-generated noise.

Ready to optimise your content for AI?

Bright Star is actively exploring innovative strategies and tools to ensure our clients remain discoverable in this evolving AI-led search environment.

If you’re looking for strategic insights, practical recommendations, or tailored solutions to improve your content’s visibility in AI searches, contact Bright Star today.

Want to ensure your brand remains visible, relevant, and ready for the future?

Let’s talk.

FAQs

What are Large Language Models (LLMs)?

Large Language Models are advanced AI systems, such as ChatGPT, that can understand, generate, and respond to human language, significantly changing how people search for information online.

How do LLMs affect traditional SEO?

LLMs prioritise delivering direct, comprehensive answers rather than lists of links. This diminishes the effectiveness of traditional keyword-based SEO, making question-answering content more valuable.

Can AI-generated content outperform human-written content?

AI-generated content can scale efficiently, but high-quality, human-written content remains essential due to its ability to build trust, provide original insights, and genuinely engage audiences.

How can businesses adapt to AI-driven search?

Businesses should focus on clearly answering audience queries, leverage content intelligence tools to understand search behaviours, and ensure their content remains authentic and relevant.

Who can benefit from Generative Engine Optimisation (GEO)?

All businesses, especially niche or B2B brands, can significantly benefit from GEO by ensuring their content remains optimised and visible in the rapidly evolving AI-driven search landscape.

Do LLMs differentiate between AI-generated and human-written content?

According to insights shared by Tom, LLMs generally do not differentiate significantly between AI-generated and human-written content. However, human-written content tends to build greater trust and authenticity, which can lead to better audience engagement.

What is the “zero-click” search phenomenon?

“Zero-click” search refers to users obtaining the information they need directly from LLM-generated answers, without clicking through to the original website. This emphasises the need for clear, concise, and directly relevant content.

Should content creators develop more FAQs on their websites?

Yes. Tom recommends clearly listing and answering common questions your audience might have. Creating comprehensive FAQs can improve your visibility in AI-driven search results by directly addressing user queries.

Are traditional websites going to become obsolete because of LLMs?

Not immediately, but websites will need to evolve. Tom suggests websites may shift towards text-focused, AI-optimised structures, though this transition depends greatly on broader human adoption of LLM-based search methods.