AI Search Optimisation for Businesses in New Zealand and Worldwide
When someone asks ChatGPT, Perplexity or Google's AI for the best provider of what you sell, the answer names a handful of businesses. My work is making sure AI assistants can find your business, understand exactly what it does, and have good reasons to include it. I have worked in search since 2006, for clients in New Zealand, the UK, Canada and the US.
What AI search optimisation is
AI search optimisation is the work of making a business easy for AI assistants to find, understand and cite when people ask them questions. It covers ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini, Claude and Microsoft Copilot. The industry also calls it AI SEO, generative engine optimisation (GEO) or answer engine optimisation (AEO). The names differ, and the job is mostly the same.
AI assistants don't show ten links. They write one answer, assembled from several sources: pages their search systems retrieve, what the underlying model learned in training, and what other sites say about you. Whether your business makes it into that answer depends on whether the assistant can read your pages, how clearly they explain what you do, and how consistently you're described across the web. I explain how AEO, GEO and SEO differ in a separate guide.
In New Zealand, Google already shows an AI Overview on 74% of commercial searches, and on 96% of advice and pricing searches, according to my NZ Google AI Overviews Study 2026.
It's not a trick or a hack. It's structured, deliberate work across your content, your site architecture and your presence on the wider web.
Platforms covered
- ChatGPT & ChatGPT Search
- Google AI Overviews & AI Mode
- Perplexity
- Gemini
- Claude
- Microsoft Copilot / Bing
AI SEO has two meanings
"AI SEO" is used for two different things. Sometimes it means using AI tools such as ChatGPT to do SEO work faster: keyword research, audits, first drafts. More often now it means optimising a business so AI search systems mention and cite it. The second is what this page is about. Some people call it GEO or AEO. Google calls it SEO.
Using AI for SEO work is genuinely useful for research, outlines and analysis, but it isn't a shortcut to rankings. Google's guidance on generative AI content says it judges pages on quality and helpfulness, however they were produced, and that mass-producing pages mainly to manipulate rankings breaks its spam policies. AI speeds up the work. It doesn't replace judgement about what your buyers need.
You'll also see LLM optimisation, AI visibility optimisation and "SEO for AI". They all describe the same shift, and no standards body defines any of them. Judge a provider by what they actually do and how they measure it, not by the acronym. My SEO and AI search glossary defines each of these terms in plain English.
How AI assistants decide which businesses to mention
For questions about products and providers, AI assistants usually search first and write second. Google describes its AI Overviews and AI Mode as built on retrieval-augmented generation, or grounding: its ranking systems fetch relevant pages from the index, and the model writes an answer based on them, with links. It also runs several related searches at once, which it calls query fan-out. What the model learned in training fills the gaps.
Retrieval: can the assistant read you?
If a crawler can't read your page, you aren't in the pool. OpenAI says ChatGPT search uses its OAI-SearchBot crawler, which is separate from GPTBot, the crawler it uses for model training, and each is controlled separately in robots.txt. Perplexity uses PerplexityBot to surface and link websites in its answers. For Google, a page has to be indexed and eligible to show with a snippet. Google says there are no additional requirements.
Understanding: is your business an unambiguous entity?
Search engines and large language models resolve a business to an entity: a named thing with a type, a location and relationships to other things, much like the knowledge graph Google uses to connect facts. Consistent facts on your site, your Google Business Profile and the wider web make that resolution less ambiguous, and so does schema markup in JSON-LD. To be plain about it, Google says its AI features need no special schema. I use structured data for entity clarity and for AI systems other than Google, not as a switch that turns on citations.
Selection: why one source and not another?
An AI answer usually draws on several sources, and the set changes between asks. Research published in 2023 found that sources which cite credible references, add statistics or include quotations were used more, while keyword stuffing did worse than no change at all. I've written up what the research on generative engine optimisation found, including its limits. Language models can also hallucinate, stating wrong things with confidence, so clear facts that are confirmed in more than one place leave less room for error about your business.
What AI search looks like in New Zealand: my research
In September 2026 I published a study of 1,000 commercially important Google searches in New Zealand. AI Overviews appeared on 73.9% of them, on 96.5% of advice searches and 96.0% of cost searches, but on only 16.5% of searches for a local provider. When an AI Overview appeared, it was the first thing on the page 88.4% of the time.
The answers lean on New Zealand sources: 80% of AI Overview citations went to .nz domains. The most cited sites were moneyhub.co.nz, a money comparison site (cited in 19.2% of AI Overviews), YouTube (18.1%) and Reddit (15.2%), and comparison and consumer-advice publishers were cited more often than the businesses they compare. People Also Ask appeared on 92.1% of searches, and not one of the 1,000 showed a classic featured snippet.
The caveat matters. This is a balanced sample of commercial searches, captured on desktop on a single day, 8 September 2026, not a share of all New Zealand searches: weighting the 275 searches with measurable Google Ads volume gave 27.0%. The full method is in how the study was built.
A second study, The Thin Market Playbook, collected 204 answers from an AI assistant with live web search across 36 industries in four countries. Of 672 websites it cited, only 19 appeared in more than one industry: for business buying questions it read providers' own websites, not directories. Asked the same question 5 times, six business industries produced an average of 16.3 websites, and none appeared in all 5 answers. In business markets the answer hasn't settled, so there is room to get in, and your own site is the main lever.
How to optimise for AI search
Make sure AI crawlers and Google can read your pages, answer the real questions your buyers ask in plain text, state your facts as facts, keep your business described the same way everywhere, and earn genuine mentions from sources your industry trusts. Then measure repeatedly, because AI answers change between asks. None of it is a hack, and most of it is good SEO done properly.
Let the right crawlers in
Check robots.txt as it is actually served, not just the file on your server. Some CDN and security settings add their own rules: Cloudflare's managed robots.txt, for example, can place Disallow rules for AI crawlers above yours. Allow the crawlers you want reading you, such as Googlebot, Bingbot, OAI-SearchBot and PerplexityBot.
Be indexable and snippet-eligible in Google
That is Google's only technical requirement for AI Overviews and AI Mode. If a page can't appear in ordinary results with a snippet, it can't be a supporting link in an AI answer either. Search Console shows which of your pages are indexed and why others aren't.
Answer buying questions on your own site
Cost, process, timeframes, what goes wrong and who it suits. My NZ study found cost and advice searches showed an AI Overview about 96% of the time. For an online store, that means category and product pages that answer real questions, not just a product grid. Before you write, check how much search demand your industry has.
State facts plainly, in text
Services, areas served, sectors, years operating, credentials and, where you can, prices. Put them on the page as text, not inside PDFs or images, so an AI system can lift them without guessing.
Make the entity unambiguous
Use the same business name, description and details on your website, Google Business Profile, LinkedIn and the directories that matter in your sector, and describe the business with Organization or ProfessionalService schema and sameAs links. Google says none of this is required for its AI features. It still removes doubt about who you are for every other system.
Earn genuine corroboration
Industry publications, trade bodies, analysts and real customer reviews. Google warns that chasing inauthentic mentions doesn't help. Original research is the strongest form of corroboration, which is why this site publishes its own.
What you don't need to do
You don't need an llms.txt file to appear in Google's AI features, you don't need to chop pages into tiny chunks, you don't need a special schema type, and you don't need to rewrite everything "for AI". Google says so in its own guide to optimising for generative AI features. Those tactics sell well. They aren't where results come from.
An honest disclosure: this site publishes an llms.txt file, because some AI systems other than Google may read it and it costs nothing to maintain. Google says it ignores it, and I'd rather tell you that than sell it to you. The same goes for schema markup. I use it because it removes ambiguity about who a business is, not because it gets anyone into an AI Overview.
What's included
Every engagement starts with a baseline of what AI assistants say about you today, then works through the foundations (crawl access, entity and schema), the content buyers need, and the third-party corroboration that makes a recommendation credible. I track it monthly and adjust as the models change, so you can see what the work is doing in plain numbers.
AI Visibility Baseline
Testing the real buying questions your customers ask across every major AI platform, asked several times over several weeks, because one check is one draw. I record mentions, sentiment and which competitors get recommended.
Entity & Schema Foundation
Structured data, entity consistency and site architecture that make it unambiguous to AI models who you are, what you do and where you do it. For businesses that serve a place, that starts with local SEO. Underneath it all sits technical SEO: checking that AI crawlers can reach your pages at all.
Answer-Ready Content
Content written to be quoted: direct answers to buying questions, specific facts and named sources, backed by genuine expertise.
Citation & Authority Building
Genuine third-party mentions from industry publications, trade bodies, analysts and real customer reviews. My research found AI assistants answering business questions read providers' own sites and industry analysts, not directories, so that is where the effort goes.
Monthly AI Visibility Tracking
Ongoing monitoring of AI mentions and AI Overview presence, including Search Console's generative AI performance report, with reporting that shows movement in plain numbers, not vanity metrics.
Continuous Adaptation
AI search changes monthly. As models update and citation patterns shift, your strategy adjusts. Twenty years in search is largely the ability to tell a real shift from noise.
How AI search results are measured
Measure AI visibility three ways: Google Search Console's generative AI performance report, which shows impressions in AI Overviews and AI Mode by page, country and device; repeated tests of real buying questions across ChatGPT, Perplexity, Gemini and Google; and your own analytics and enquiries. Rankings alone no longer tell you whether you're being seen, and a single AI answer tells you almost nothing.
Google launched the generative AI performance report on 3 June 2026, and it has been available to all sites since 31 August 2026. Google also says no third-party tool has access to its internal ranking or AI systems, so treat any tool's "AI visibility score" as an estimate, not a verdict.
Repetition matters because AI answers vary. When my NZ study re-ran 100 searches, AI Overview presence agreed 93% of the time. In the Thin Market Playbook, no website appeared in all five answers when the same business question was asked five times. One check is one draw.
What nobody can promise you
No one can guarantee that ChatGPT, Google or any other AI assistant will mention your business. The models change, answers vary between asks, and the providers decide what to show. What I can do is remove the reasons they'd leave you out, and show you, in plain numbers, whether it's working.
Google's own documentation says indexing and serving isn't guaranteed, even for a page that meets every requirement. Be wary of anyone who promises citations in ChatGPT, a place in AI Overviews or results by a fixed date. My client work has been SEO for most of two decades and is now moving into AI search, and clients judge it the same way they always have: on results.
Marcos has done an exceptional job for us and always focused on delivering results not just spending our money. With so many options out there we are very grateful to be able to have a partner like Marcos who we can trust and understands our business.
Go deeper: GEO, AEO and the future of SEO
Three guides go further than a service page can. They cover where the term GEO came from and what the research really found, how AEO, GEO and SEO compare, and whether SEO is still worth paying for now that AI writes the answer.
What is generative engine optimisation (GEO)?
Where the term came from, what the 2023 research found, and what it can't tell you.
What is GEO?AEO vs GEO vs SEO
What answer engine optimisation is, and how the three compare side by side.
AEO vs GEO vs SEOWill AI replace SEO?
Whether SEO is still worth it in 2026, and what ChatGPT can and can't do for it.
Will AI replace SEO?AI search optimisation FAQs
What is AI search optimisation?
AI search optimisation is the work of making your business easy for AI assistants such as ChatGPT, Google AI Overviews, Perplexity and Gemini to find, understand and cite. It's also called AI SEO, GEO (generative engine optimisation) or AEO (answer engine optimisation). Instead of competing only for a ranking position, you're improving the odds that AI models include and recommend your business when people ask about what you sell.
How is it different from SEO?
Traditional SEO targets ranked lists of links. AI search optimisation targets the single answer an assistant writes. The foundations are the same, and Google says optimising for its AI features is still SEO. What changes is the emphasis: clear entity information, answers that can be quoted, genuine third-party corroboration, and measuring mentions as well as rankings. Strong SEO helps AI visibility, but it isn't the whole job.
How do you optimise for AI search?
Start by making sure AI crawlers and Google can read your pages. Then answer the questions your buyers actually ask, in plain text on your own site, keep your business facts consistent everywhere, add accurate schema markup, and earn mentions from sources your industry trusts. Test the same buying questions across several assistants over several weeks, because answers vary between asks.
What is the best AI for search engine optimisation?
There isn't one best AI. General assistants such as ChatGPT, Claude and Gemini are good for research, outlines and first drafts. SEO platforms add data an assistant doesn't have, such as search volumes and rankings. None of them replaces judgement about what your buyers need. Google says no third-party tool has access to its internal ranking or AI systems, so treat any tool's "AI score" as a guide, not a verdict.
What is the 30% rule for AI?
There's no official "30% rule" for AI content from Google or any other search engine. The phrase circulates online with different meanings. Google's published position is that it judges content on quality and helpfulness, not on how much of it a machine wrote, and that producing content mainly to manipulate rankings breaks its spam policies. Use AI where it helps, and check every fact yourself.
Is SEO still worth it in 2026, or do I need AI search instead?
Yes, SEO is still worth it, and you don't have to choose. Google's AI features are built on its core search systems, so the same foundations that earn rankings also make you eligible to be cited. What has changed is what good SEO has to cover, and I've set out the evidence in is SEO still worth it in 2026. For the rankings side, see SEO in Auckland or SEO across New Zealand.
What is replacing SEO?
Nothing is replacing it outright. Search is gaining a new layer: AI answers that sit above or alongside the links. In my study of 1,000 New Zealand commercial searches, AI Overviews appeared on 73.9%, and not one search showed a classic featured snippet. The job has widened from ranking pages to being the source those answers are built from.
Can I do SEO myself?
Yes, many businesses can do the basics themselves: a crawlable site, clear pages that answer buyers' questions, accurate business details and genuine reviews. Google itself says a small local business can probably do much of the basic work on its own. Where specialist help pays is diagnosis, prioritisation, and the technical and authority work that takes experience. The free audit shows which parts you can safely keep in-house.
How do I know if AI models mention my business?
Ask the buying questions your customers ask, across ChatGPT, Perplexity, Gemini and Google, several times over a few weeks, and record whether you're named, how you're described and who's recommended instead. Add Search Console's generative AI performance report for Google's side. That's what my free AI Visibility Audit does, for businesses in New Zealand and overseas.
How long does AI search optimisation take to work?
Fixes to crawl access, indexing and on-page answers can show up as soon as the pages are recrawled, which can take days or weeks. Building the authority and third-party corroboration that make AI models recommend you takes longer: in my experience, typically three to six months of consistent work, similar to SEO timelines. Anyone promising faster, guaranteed AI citations is promising something they don't control.
How much does AI search optimisation cost?
AI search optimisation starts at $2,000 a month excluding GST, or it is built into an SEO retainer from $2,500 a month so both engines share one strategy. Standalone AI search programmes in New Zealand typically run $1,000 to $3,000 a month. The full breakdown is in what SEO and AI search cost in NZ.
Start with the free AI Visibility Audit
Know exactly where you stand in ChatGPT, Google AI Overviews and Perplexity before spending anything. No cost, no obligation.
Book Your Free Audit