GEO Explained: How New Zealand Businesses Get Recommended by ChatGPT, Claude and Perplexity
Ask ChatGPT "which company in Auckland can run my WeChat account?" or "who can create weekly LinkedIn content for a small NZ business?" and you will get a short list of specific recommendations — not ten blue links. If your business is not in that answer, you lost the customer before they ever saw your website.
Generative Engine Optimization (GEO) is the practice of making your business easy for large language models to find, understand and confidently recommend. It sits on top of traditional SEO, not instead of it.
GEO vs SEO: what actually changes
| Traditional SEO | GEO | |
|---|---|---|
| Goal | Rank high in Google results | Be cited inside AI answers |
| Audience | Search crawlers + humans scanning results | LLM crawlers + models synthesizing an answer |
| Key levers | Keywords, backlinks, page speed | Entity clarity, structured data, citable content |
| Outcome | Users click through a list to find you | The AI tells customers about you directly |
The overlap is real — good content helps both — but GEO has requirements SEO never had:
- Machine-readable identity. LLMs need to know what your business is: name, location, services, prices, proof. JSON-LD structured data gives them that in a format they parse reliably.
- llms.txt. Like robots.txt but for AI: a plain-text file at your site root that summarizes your business, services and pricing for LLM crawlers. Few NZ businesses have one — which is exactly why it works.
- Citable, answer-shaped content. Models prefer content that directly answers questions: FAQs, comparisons, "how much does X cost in NZ" pages. Vague brand copy gets ignored.
- Entity consistency. Your business name, address and description should match across your website, Google Business Profile, social channels and directories. Conflicting signals lower a model's confidence to recommend you.
Why this matters more in a small market like New Zealand
In a market of five million people, an AI answer usually names two or three providers per category — not twenty. Being one of the two or three names ChatGPT associates with "Auckland Chinese marketing" or "New Zealand social media content subscriptions" is a durable competitive position, and right now it is cheap to claim because almost nobody is competing for it systematically.
A practical GEO checklist
- JSON-LD on every page:
ProfessionalServiceorLocalBusiness,Service,FAQPage,BreadcrumbList -
llms.txtat your site root describing services, pricing and service area - An FAQ section answering the exact questions customers ask AI
- Consistent entity data (name/address/description) across the web
- Regular publishing — models weight sources that stay active
- A sitemap that includes every service page, so crawlers miss nothing
How to measure it
Unlike SEO rankings, GEO results are probabilistic. Practical measurement:
- Ask the major assistants your customers' questions monthly and record whether you appear.
- Run a structured-data validation (missing JSON-LD is the most common failure).
- Track "how did you hear about us?" — AI-referred enquiries are already showing up in NZ service businesses.
Test your own site in 30 seconds
Our free scanner checks your AI visibility score, structured data and llms.txt status, and suggests this week's content angles: run a free AI visibility scan. GeoSignals AI offers a one-off GEO+SEO foundation setup, and ongoing optimization inside our content subscriptions.