Our clients are asking about AI and search. We're hearing concerns about reduced visibility, dropping website traffic, confusion about new terms like AEO and GEO, and uncertainty about whether traditional SEO still matters when people are increasingly using AI tools for online research.

The reality is we're in a period of rapid change where new baselines haven't been established yet. That naturally creates confusion, but it doesn't need to be overwhelming. When we boil it down, search is a behaviour, not a channel. When you think of it like that, there's only a handful of things that matter for staying visible in this evolving world of search.

You've probably seen terms like AEO (Answer Engine Optimisation) and GEO (Generative Engine Optimisation) floating around. People use these terms differently, and there's no universal agreement on what they mean or whether they're even separate things.

The reality is that the tactics are similar regardless of which term you use. Whether you're trying to get featured in Google's AI Overviews or cited by Claude, you're optimising for the same basic qualities: clear, helpful content that directly answers people's questions.

We use the term "AEO" with our clients because it's more established. To be honest, the specific terminology matters less than understanding what actually works.

All AI systems, whether they're powering Google's search results or tools like Perplexity, want the same thing: content they can easily understand and confidently cite. It comes down to two things:

You need both parts working well. Great content that AI systems can't understand gets ignored, while perfectly marked-up content that doesn't help people leaves AI with nothing worth sharing.

The good news is that what makes your content work well with AI also strengthens your traditional SEO. You're not choosing between old and new approaches – you're building foundations that work across all the ways people search for information (another reminder that search is a behaviour not a channel!).

Even better, you can make a lot of improvements without engaging a developer or your digital agency (you're welcome).

Here are some tips from our team:

1. Think about your content differently. Make existing content more helpful and direct. When creating new pages or updating old ones, think about the questions people are actually asking and structure your content around clear answers.
2. Consider the technical side. Schema markup and structured data act as a separate layer that helps AI systems and search engines understand your content. It's not about rewriting anything, it's about labelling what's already there so machines can read it properly.
3. Stay flexible. This space is changing quickly. Focus on being genuinely helpful rather than trying to game every individual platform.
4. Don’t set and forget. Your website is a living, constantly evolving digital asset. Depending on the platform, LLMs and Google review your content every 2–12 weeks, so regularly check your analytics, search results and even perform manual searches on LLMs as a guide for continually improving your content.

About one in five Google searches now shows AI-generated summaries, and more people are using AI tools for search. This doesn't make traditional search optimisation pointless, but websites that only focus on traditional search miss opportunities to be found in the places people are now searching.

The organisations doing well focus on making their knowledge genuinely useful and easy to find. The same things that make content valuable for humans – being clear, authoritative, and helpful – also make it valuable for AI systems to discover and share.

We're laying foundations that work regardless of how technology changes. The basics of creating quality, user-focused content remain key, whether that content gets found through Google, ChatGPT, or whatever comes next.

In 2025 our team delivered a new website for Stroke Aotearoa New Zealand. The site went live, SEO was looking good — happy days. But during a monthly Site Strategy Session we noticed that one random news article was getting outsized attention.

In a nutshell: AI had read a page about Dave Matthews (a police officer from the Far North who suffered a stroke) and conflated it with Dave Matthews (the famous American musician). The result: an AI-generated news article confidently touting that renowned musician Dave Matthews had suffered a stroke, combining an image of Dave Matthews (musician) with content about Dave Matthews (Far North police officer).

Thus began the self-reinforcing loop of AI. The more people clicked on the Stroke Aotearoa article linked in LLMs and AI summaries, the more 'accurate' AI assumed it was. An American newspaper even reported the AI-generated article as fact!

Our former colleague Richie Jose has detailed the story in more depth on his personal blog.¹

Many of our smaller clients are wondering if all the AI hype is even relevant to them? In practice, we've noticed that the smaller you are, the more risk there is of being swallowed up and misinterpreted by AI without you even noticing it's happening.

Ready to navigate this evolving landscape? We'd be happy to chat about how these ideas might work for your specific situation.

And if you're feeling generous, please consider supporting the wonderful team who do such important mahi at Stroke Aotearoa NZ.

¹Richie Jose, The cop, the rockstar, and the AI algorithm: Lessons from "Dave Matthews Gate"