AI Search Optimisation for NZ Businesses: ChatGPT, Gemini and Google AI Mode

New Zealand businesses are now being found through ChatGPT, Gemini and Google AI Mode, not just ranked pages. That's not a future shift to prepare for, Google AI Mode has been live in NZ for close to a year, and it's already how a growing share of searches get answered.

Optimising for one of these doesn't automatically cover the others, each system retrieves, interprets and cites information differently, which means "AI optimisation" isn't one task it's several related ones, and most NZ businesses are only doing the traditional SEO version of it.

At AI Optimisation, we treat this as the same underlying problem across every platform: Is this business clear enough that AI can confidently reference it? This article covers what changes for ChatGPT, Gemini and Google AI Mode specifically, and where they overlap.

How AI Search Optimisation Differs Across Platforms

Answer Engine Optimisation (AEO) is about optimising for direct answers, the short, specific response a system gives when someone asks a factual or local question.

Generative Engine Optimisation (GEO) is about optimising for how generative AI models summarise and cite content when building a longer answer.

Together, AEO and GEO get your business discovered, recognised and referenced across AI-led search but ChatGPT, Gemini and Google AI Mode each apply these principles slightly differently.

ChatGPT Optimisation for New Zealand Businesses

ChatGPT rarely shows its sources, and it weighs consistency across independent sources more heavily than live ranking signals which makes it a genuinely different optimisation problem to Google. We've covered the mechanics and what actually helps in detail here.

Gemini Optimisation for New Zealand Businesses

Gemini is more tightly integrated with Google's existing index and Search Console data than ChatGPT is, which means the technical fundamentals - crawlability, structured data, page speed all carry more direct weight than they do for a model working mostly from training data.

For NZ businesses, this means getting the basics genuinely right: clean schema markup (LocalBusiness and FAQ schema especially), a site Google can crawl without JavaScript rendering issues, and content that answers specific questions directly rather than only describing services in general terms. Gemini rewards clarity it can verify against your existing Google presence, not just clarity on the page itself.

What Google AI Mode Means for NZ Businesses

Google AI Mode uses generative AI to provide answers directly inside the search results, rather than just linking out to websites. Instead of listing pages, it summarises what it considers the most reliable information and cites the sources it trusts.

Take a local mechanic as an example. A customer searches "how much does a WOF cost in Auckland." Instead of clicking through several sites, AI Mode might display a direct answer with the average price, citing one or two sources it trusts. If a mechanic's website has a clear, direct FAQ answering that exact question, their business can appear in that summary ahead of larger competitors who never wrote the answer down plainly.

That principle holds for any local business, not just mechanics: a café, a law firm, a plumber. The businesses that get cited are the ones that already wrote the direct answer somewhere AI Mode can find it.

SEO Factor	Traditional SEO	AI Mode Optimisation
Keywords	"Mechanic Auckland"	"How much does a WOF cost in Auckland?"
Content	Service pages	FAQ-style direct answers
Visibility	Ranked links on page one	Cited inside the AI answer
Schema	Optional	Essential
Authority	Backlinks	Clarity and structured context

A Practical Starting Checklist

Start with the questions your customers actually ask, not the ones you'd guess. Pull them from real enquiries, sales calls or reviews, a mechanic might focus on "how often should I service my car in NZ" or "what's included in a WOF check," while a law firm's version will look entirely different.

Write direct answers to those questions in plain, conversational language short enough that a system could lift the answer as-is, roughly 40 to 60 words. Add FAQ schema so search systems can identify those answers as trustworthy, and LocalBusiness schema so your address, hours and reviews are machine-readable.

From there, build supporting content around the same topics like a "top 5 signs you need brake repairs" piece linking back to a core service page does more for topical authority than a dozen generic posts. Keep the on-page basics tight: specific page titles, descriptive alt text, and internal links that connect related questions to the pages that answer them.

Frequently Asked Questions

Will AI Mode replace SEO for NZ businesses? No. Rankings still matter, they're part of how content gets retrieved in the first place but ranking alone no longer guarantees you get cited in the answer itself.

How do I know if I'm showing up in AI Mode or being cited by ChatGPT and Gemini? Track impressions in Google Search Console for AI Mode specifically. For ChatGPT and Gemini, the most reliable method is direct: ask them the real questions your customers ask, regularly, and check whether your business appears.

Do I need separate strategies for ChatGPT, Gemini and AI Mode? Not entirely separate, the foundation (clear entity information, direct answers, structured data, independent validation) is the same across all three. What changes is which of those matters most for each: consistency for ChatGPT, technical fundamentals for Gemini, direct-answer content for AI Mode.

What's the difference between AEO and GEO? AEO is optimising for direct answers. GEO is optimising for how generative models summarise and cite your content more broadly. Most businesses need both.

This is the same underlying work across every platform: entity clarity, structured content, independent authority, and content that directly aligns with what people are actually asking, the five layers the AI Visibility Engine™ is built around. Getting found by AI search isn't really five separate jobs for five separate platforms. It's one job, evaluated five different ways.

AI can't recommend what it doesn't understand.