For more than two decades, we have trained people to search the web in keywords. 

“Health insurance family.” “Council waste collection.” “Super contribution limits.” “Airport parking Sydney.” 

But that behaviour is changing rapidly. 

The rise of ChatGPT and other Large Language Models, combined with the introduction of AI-powered experiences such as Google AI Mode, is teaching people that they no longer need to translate what they want into a handful of search terms. They can simply ask the question they actually have. 

Instead of searching for “health insurance family”, they might ask: 

“What should I consider when choosing private health insurance for a family with two young children?” 

Instead of “council waste collection”, it becomes: 

“I’ve just moved into the area. When is my recycling collected and how do I get another bin?” 

This isn't simply a change to Google Search. It represents a much broader change in how people expect to find information. 

And that has significant implications for your own website. 

Search is becoming conversational 

Google's own data provides a useful indication of the scale of the change. The average query in AI Mode is now around three times longer than a traditional Google search. People are increasingly using Search for more complex questions, planning, comparisons and decision-making, and continuing their searches through follow-up questions. 

The search box is becoming less like an index and more like the beginning of a conversation. 

This creates an interesting disconnect for many organisations. 

A customer can now ask Google, ChatGPT or another AI platform a detailed question about your organisation, products or services in completely natural language. But once they arrive at your website, they can often still be confronted by a search experience designed around the keyword behaviours of 10 or 15 years ago. 

The expectation has changed. The experience hasn't. 

Your external and internal search strategies are converging 

Until recently, SEO and on-site search could largely be treated as separate disciplines. 

One helped people find your website. The other helped them navigate it once they arrived. 

AI is starting to break down that distinction. 

Increasingly, both depend upon your organisation's ability to understand the intent behind a question and connect that intent to authoritative, well-structured content. 

This means organisations need to think about search as an interconnected ecosystem. 

Your content needs to be discoverable by Google. It needs to be understandable and citable by LLMs. And when someone arrives at your digital experience, your own search capability needs to understand the same natural-language questions they have become accustomed to asking elsewhere. 

If it can't, the experience can very quickly feel outdated. 

The content itself needs to change 

This isn't a problem that can be solved simply by replacing the search box. 

Conversational search is ultimately a content challenge as much as a technology challenge. 

Organisations should begin by looking at the questions their audiences are asking. 

Google Search Console provides one source of intelligence. Your existing on-site search logs provide another. AI visibility monitoring can increasingly reveal the prompts and topics through which brands are being discovered and referenced by LLMs. 

Together, these sources provide something incredibly valuable: a growing picture of audience intent. 

Rather than structuring content primarily around the terminology used internally by the organisation, teams can start structuring it around the questions people need answered. 

FAQ content is an obvious place to start, but the principle should extend much further. 

At a page level, organisations should consider the likely questions that bring someone to that content. Does the page answer them clearly? Is the answer buried within hundreds of words of introductory copy? Are related questions addressed? Is important information expressed consistently elsewhere on the site? 

The objective isn't to stuff pages with artificially constructed questions. It is to create content that provides clear, authoritative answers to genuine user needs. 

Structure matters too 

Making content understandable to machines is becoming almost as important as making it understandable to people. 

That means reviewing the underlying structure and metadata surrounding your content. 

Schema and other forms of structured data can help search engines understand what a piece of content represents, whether it is a product, service, organisation, event, location, FAQ or something else. 

Clear information architecture, descriptive headings, structured content models, consistent terminology and meaningful relationships between content all contribute to making information easier to retrieve and interpret. 

These practices aren't exclusively about GEO, SEO or on-site search. Increasingly, they support all three. 

Measure what people are asking, not just where they clicked 

For years, digital analytics has concentrated heavily on pages, sessions, clicks and conversions. 

Conversational search creates another valuable dataset: questions. 

What are people asking Google before they reach you? What are they asking your website once they arrive? What questions are people asking AI platforms about your category? Where is your organisation being cited? Where are competitors appearing instead? And what questions aren't being answered adequately by your existing content? 

Search Console data, on-site search analytics and AI visibility monitoring should increasingly be considered together. 

This creates a continuous optimisation loop. 

Listen to the questions. Identify gaps. Improve the content. Measure whether visibility and answer quality improve. Repeat. 

Your search technology may need to change as well 

There is, however, a point at which better content can't compensate for outdated search technology. 

Traditional site search has generally been designed to return a ranked list of pages matching a collection of keywords. 

Increasingly, users expect something different. 

They expect to ask a question, have the system understand its meaning and context, receive an appropriate answer, see where that answer came from, and then refine their question without starting again. 

For organisations with large or complex digital estates, this makes the search capability of the underlying DXP an increasingly important consideration. 

Platforms such as Squiz Conversational Search are bringing conversational discovery directly into enterprise digital experiences. Other composable technologies, such as Wizen, can provide AI-assisted answers grounded in an organisation's approved content, with the capability embedded into an existing website, portal or application. 

This also creates an opportunity to decouple search from the DXP itself. 

Rather than accepting whichever search capability happens to come bundled with a CMS or DXP, organisations can increasingly adopt a composable approach, selecting a search and retrieval layer appropriate to their audiences, content, governance and risk requirements. 

A practical roadmap for the new search environment 

For organisations considering how prepared they are, the starting point doesn't need to be a major technology transformation. 

Start with the questions. 

Understand how people currently discover your organisation through traditional search and AI. Analyse what they search for when they reach your website. Identify the high-value questions and intents that matter most to your audiences. 

Then examine whether your content actually answers them. 

Improve content structure and Schema. Rewrite and expand content where important questions aren't being answered. Establish AI visibility monitoring alongside Search Console and on-site search analytics. And assess whether your existing search technology can understand natural-language intent rather than simply matching keywords. 

Only then should technology decisions follow. 

Because the fundamental change isn't really about search technology. 

It's about search behaviour. 

People are rapidly becoming accustomed to simply explaining what they want and expecting technology to understand them. 

The organisations that adapt will make their information easier to find, understand and use, wherever the question is being asked. 

Those that don't risk providing a digital experience that increasingly feels like it belongs to an earlier era of search.