Writing for AI search: practical ways to boost AI visibility and engage users

AI search is now a huge part of how people research and shop, with ChatGPT alone capturing a whopping 900 million weekly active users.

It’s no surprise then that the marketing world is swamped with AI-centric conversations. Just scrolling through your LinkedIn feed, chances are you’ll find five or six posts with ‘hacks’ to conquer AI search. How do we not get lost amongst the noise of new tracking tools or content techniques that promise to crack the system at the expense of content quality? 

The truth is, we need to rely on the tried-and-tested tactics that help content surface in LLMs, building them into content strategy to meet our audiences where they’re searching, and (best of all) making sure that content is still actually quite nice to read if you’re a real person, and not an algorithm.

Why write for AI search?

Good content has always aimed to be useful, clear and authoritative; writing for AI search simply makes those qualities even more important. As we’re all grimly aware by now, the introduction of AI search has led to plummeting click-through rates; top-ranking organic brands have dropped by roughly 34% to 58% where AI Overviews are present. To make brands not just show up, but stand out in AI search, we need to adjust our approach to make sure our content is showing up in the places people are searching, questioning and comparing. 

What does it mean to write for AI search?

LLMs need to understand the meaning, context and relevance of content, not just to match exact keywords. LLMs parse content at a deeper level, considering the relationships between topics, concepts and information in each individual passage. As a result, a page can surface in the results even if it doesn’t use the exact phrasing in the query, provided its content has a clear connection to the subject. 

This means thinking beyond keyword placement, focusing instead on creating copy that has clear meaning and clear context, where the relationship between topics is made explicit.  

This isn’t exactly radical: content should always be structured so that useful information is easy to identify, understand and extract. What’s changed is the incentive to make our content visible, well, everywhere. The objective was never to write purely for an algorithm; it’s to make valuable information clear, useful and accessible. And doing that well now also happens to serve GEO.

Writing content for AI search: Core techniques

The way information is structured can influence how effectively it answers specific search queries and how individual passages are understood in context. These approaches help create content that can address a broader range of queries without relying on keyword stuffing or making your content unbearable to read once it’s in front of your audience.

Utility writing

A utility writing approach gives the reader the answer, definition, recommendation or key information they are looking for before providing the detail.  

For example, a page targeting the question “What is MVHR?” could start with a straightforward definition: “MVHR, or mechanical ventilation with heat recovery, is a whole-house ventilation system that extracts stale air and supplies fresh air while recovering heat from the outgoing air.” The copy can then explain how the system works, its benefits, and where it is most suitable. 

Each sentence should survive on its own. The evidence shows that proposition-level retrieval (where systems decompose text into atomic, self-contained claims) outperforms passage- and sentence-level retrieval. The tricky part is applying this style when it’s more than just facts. Because content is often storytelling, and getting the reader to feel something too.

Structuring content

Across the page, content structure can also have a significant effect on how AI systems cite copy. 44% of ChatGPT citations come from the first third of content, so if you want to optimise for AI search, you need to make sure those key queries and connecting questions are answered quickly and succinctly. 

At paragraph level, 53% of citations come from the middle of paragraphs. This is because AI models have been trained using journalistic and academic writing, with some contextual framing before the main point. 

Where traditional search used to reward depth, AI search wants clear answers up front. So, when it comes to structuring your copy for AI and Google search optimisation, it’s best practice to front-load key insights. Focus on keeping paragraphs clear and informed, rather than shoehorning dense info into the first sentence.

Building connections between entities

Identifying key entities and their associations is the secret to creating interlinks between related queries in your content. For example, the entities in a blog about home ventilation might look something like this:

• Places: Council homes, blocks of flats, apartments, houses 
• People: Home builders, landlords, homeowners, tenants, building managers
• Cultural concepts: Indoor air quality, energy efficiency, airtightness, damp and mould 
• Related entities: Mechanical ventilation, MVHR, extractor fans, trickle vents, Part F, ventilation rates

Using descriptive headings, contextual references, and relevant internal links means LLMs like ChatGPT or Claude can understand how each entity relates to another, and how the page itself fits within the context of the wider topic. 

Chunking information into useful sections

Chunking breaks down complex information into clear, self-contained sections. Each section will have a clear purpose and is designed to answer a specific user need. To ensure we answer that question as effectively (and efficiently) as possible, we can adapt our copy’s format to best match the user intent. 

For example, we could combine clear, descriptive H2s and H3s with short paragraphs to succinctly deliver information. However, for a more comparative search, such as “Which types of probiotics are the best for bloating?”, copy would be more digestible in a table format. Similarly, a checklist set of requirements is better presented in a bullet point list, rather than a lengthy paragraph. This makes long-form content easier for humans to navigate, while helping AI systems identify individual pieces of relevant information.

Providing a unique point of view

As copywriters, our job isn’t just to repackage information that already exists online. AI systems can draw on a wide range of sources, so adding a genuine point of view gives your content something more distinctive to work with. Think about the experience, expertise or perspective you can bring to the topic. Ask yourself, can you:

• Offer a unique perspective? 
• Include an expert comment? 
• Summarise a practical use case? 
• Provide a real-world example or case study?

This gives content some originality that generic summaries can’t easily replicate. It also gives AI more useful, specific information to draw on when answering related queries, while still giving readers a reason to engage with your content rather than looking elsewhere.

What to avoid if you want to stay visible in AI search

For all the content techniques you should use, there’s plenty you shouldn’t. Stick to these top tips:

• Don’t use long walls of text: Ideas can too easily blur together and leave entities indistinct, making it harder for AI to separate the content into digestible sections. 
• Don’t hide important answers: Burying your answers in expandable menus or swanky design features may look nice, but AI can struggle to pick up the content, meaning key information is skipped in its answers 
• Don’t get addicted to PDFs: Search engines will index text-based PDFs, but they lack the structure and metadata that normal pages provide. Keep key product information on product pages – not hidden in brochures. 
• Don’t lump key info in with images: AI systems can extract and interpret text from images, but the extra work to do so leads to the types of inaccurate answers AI is famous for. If you include a visual, make sure you provide alt text, or present critical details as copy elsewhere on the page.

AI search techniques that strengthen SEO best practices

The principles associated with AI-search optimisation will already be familiar to anyone working in SEO or content. The difference is that we are becoming more conscious of how these practices affect the way information can be interpreted and retrieved by AI systems:

Keywords and query intent

Keyword research is still an important part of creating content, but a single keyword doesn’t tell us everything a user wants to know. Looking at the questions and related searches reveals the wider intent behind a query, like definitions, practical questions, comparisons, problems, recommendations or questions that surface later in the user journey. With a query fan-out approach, we can ensure content addresses a subject and any supporting queries in full, rather than simply repeating a target phrase or keyword.

Readability and retrievability

Shorter paragraphs, clear language, logical structure and descriptive headings make information easier for people to understand and navigate. Retrievability builds on this by ensuring key information is clearly structured and easy to identify, making it easier for both users and AI systems to find and reference the relevant copy. 

Not every answer needs to be reduced to a short snippet: this can easily become an exercise in over-optimised sentences which are too self-contained to be readable, let alone enjoyable. Complex topics will still need depth, evidence and context. Instead, the copy should make that depth easier to navigate by giving each piece of information a clear purpose and place.

The winning formula is alignment with user intent, comparison against alternatives, inclusion of external evidence like reviews, and preserved factual accuracy.

Using AI-search principles to shape content for real people

Understanding the questions people ask helps us to identify opportunities to answer these questions, creating a more connected approach between informational and commercial intent. In AI search, a user’s search journey rarely ends with their first question. Someone searching for a product is likely to expand beyond a single query, using their initial question as a jump pad for other informational searches:

What is it? 

↓

How does it work?

↓

Is it right for me? 

↓

What are the alternatives? 

↓

What should I know before buying?

When writing with AI search in mind, we can anticipate these questions, whether employing blogs to address broader informational queries or adapting and updating product pages to ensure information is accessible. Can product information be made more scannable? Are there FAQs to answer common questions? Do detailed descriptions provide the context users are looking for, whilst still being retrievable for AI models? 

The future of copywriting is better content, not AI-only content

Writing for AI search doesn’t mean copywriters and content strategists must abandon the practices that have always made content effective. What is changing is the ways that content can be discovered and consumed. Sure, AI search makes clarity, structure, usefulness and retrievability even more important, but these are the qualities we should already be aiming for in our copy. 

Techniques for writing for AI should be part of the everyday, to improve visibility in LLMs and AIOs, and to make content better for the people reading it. 

The rules around AI continue to change, so for now, the strongest approach I can recommend is a simple one: aim to create genuinely useful, authoritative content, and structure it so that its value is easy to understand, navigate and retrieve across all search journeys, wherever they start. 

Need help getting your content in front of those 900 million weekly ChatGPT users? At Distinctly, our content writing services can help your content get retrieved and referenced, while our GEO and SEO expertise makes sure the same posts and pages continue to perform on Google and LLMs.