A buyer types a question into ChatGPT. Not a search query - a question. Something like "what's the best B2B SEO agency in New Zealand for SaaS companies?" The AI responds with a recommendation. Your competitor gets named. You don't. This isn't a Google ranking problem. This is an AEO problem - and fixing it requires a completely different playbook.
SEO vs AEO: the actual difference
SEO - search engine optimisation - is the practice of making your site rank higher in Google's results. You optimise for a crawler that indexes pages, scores signals like backlinks and page speed, and returns a ranked list of URLs for a human to click through.
AEO - answer engine optimisation - is the practice of making your content the source an AI cites when it generates a response. You're no longer optimising for a ranked list. You're optimising to be the answer. The AI reads your content, decides whether it trusts it, and either quotes you, paraphrases you, or ignores you entirely.
That distinction changes almost everything about what good work looks like. The goal isn't to rank above a competitor - it's to be the source an AI reaches for when a question lands in your category.
"SEO is about getting a human to click your link. AEO is about getting an AI to trust your content enough to repeat it."
How AI engines actually read your site
When a large language model is trained or updated, it ingests vast amounts of web content. When it's responding to a query, it draws on that training plus - in the case of tools like Perplexity, ChatGPT with search, or Google's AI Overviews - live retrieval from the web. Understanding both layers matters.
The training data layer
During training, AI models prioritise content that is clear, authoritative, structured, and unambiguous. They're pattern-matching for "what is the most reliable answer to this type of question." Content that hedges everything, buries the point, or reads like it was written to rank rather than inform gets deprioritised - or ignored entirely.
This is why most B2B content fails the AEO test before it even gets to technical setup. The writing itself isn't built to be cited. It's built to be found. Those are different goals, and they produce different content.
The retrieval layer
When AI tools pull live content at query time, they're looking for specific signals: schema markup that describes what your content is about, clean page structure that makes the hierarchy clear, and content that directly and specifically answers the type of question being asked.
A page that ranks well in Google can still score zero in live AI retrieval if it doesn't have the right schema, if the content structure is ambiguous, or if the answer to the question is buried three paragraphs down under a H2 that doesn't signal what the section contains.
What AEO work actually looks like
This is where most "AEO guides" fall apart. They describe the goal without describing the work. Here's what AEO implementation actually involves, in the order it makes sense to do it.
Structured data and schema
Schema markup is machine-readable metadata that tells AI engines - and Google - what your content actually is. An FAQ schema tells an AI that this page contains questions and answers. An Organisation schema tells it who you are, what you do, and whether you're a real business. A Product schema tells it exactly what your product does and who it's for.
Most B2B sites have either no schema or outdated schema from a plugin that was set up in 2019 and never touched since. The gap between "has some schema" and "has correctly structured, comprehensive schema" is where most of the AEO opportunity sits.
// JSON-LD in your page <head> { "@context": "https://schema.org", "@type": "FAQPage", "mainEntity": [{ "@type": "Question", "name": "What is AEO?", "acceptedAnswer": { "@type": "Answer", "text": "AEO (Answer Engine Optimisation) is the practice of structuring content so AI tools cite it in responses..." } }] }
llms.txt
The llms.txt file is the robots.txt of the AI era. It's a plain-text file you place at the root of your domain that tells AI crawlers what your site is about, which pages are most important, what your product does, and who your ideal customer is - in plain language the AI can use directly.
Adoption is still low. Most NZ B2B sites don't have one. That's a fast-win moat while it lasts - in 12 months, not having one will be the equivalent of not having a sitemap.
Content structure and directness
AI engines reward direct answers. If someone asks "what does [your product] do," your homepage should answer that question in the first two sentences - not after a hero section, a value proposition carousel, and three testimonials.
The structural fix is about hierarchy and specificity. Every H2 should signal clearly what the section answers. Every section should open with the answer, not build toward it. The inverted pyramid that journalists have used for a century turns out to be exactly right for AEO too.
- Lead with the answer: state it in the first sentence of every section, not the last
- Use specific H2s and H3s: "How pricing works" beats "Pricing" every time
- Write for the question: frame sections around the question a buyer would actually ask
- Avoid filler openers: "In today's fast-paced digital landscape" tells an AI exactly nothing
Entity authority and brand mentions
AI engines build a model of entities - companies, products, people, concepts - and their relationships. The more consistently your brand, product name, and positioning appear across the web in a coherent way, the stronger your entity authority becomes. This means your About page, your LinkedIn company description, your PR mentions, and your partner listings all need to say the same things in consistent language.
Inconsistency confuses the model. If your homepage says you're a "B2B SaaS platform," your LinkedIn says "enterprise software," and your Crunchbase says "technology startup," an AI has to pick one or none. Entity consolidation is unglamorous work that almost nobody does, and it moves the needle significantly.
Why most B2B sites fail AEO
It's not that B2B marketers are lazy or unaware. It's that most B2B content was built for a fundamentally different purpose - ranking on Google and converting humans who clicked through. The signals that matter for AI citation are different enough that even a technically strong SEO site can be invisible to AI engines.
| Signal | Matters for Google | Matters for AI citation |
|---|---|---|
| Backlinks | ✓ High weight | ~ Some weight |
| Page speed | ✓ Ranking factor | ~ Crawlability only |
| Structured schema | ✓ Rich results | ✓ Critical |
| llms.txt | ✗ Not relevant | ✓ Critical |
| Direct answers in copy | ~ Featured snippets | ✓ Critical |
| Entity consistency | ~ Knowledge graph | ✓ High weight |
| Keyword density | ✓ Still relevant | ✗ Irrelevant |
The most common failure mode is a site that has good SEO fundamentals - clean URLs, fast load times, reasonable backlink profile - but content that was written to rank a keyword, not to answer a question. The H2s are keyword phrases. The intro paragraphs are padded for length. The actual answer to "what does this product do and for whom" takes four paragraphs to arrive at.
"Most B2B content was written to tell Google the page is about a topic. AEO requires content written to tell a buyer the answer to a question. They look similar on the surface. They're structurally completely different."
The NZ B2B gap
New Zealand B2B companies are behind the curve on AEO - not because the market is unsophisticated, but because most NZ marketing teams are lean, and AEO has only become commercially urgent in the last 18 months. The positive read on this is that the window for competitive advantage is still open.
A NZ SaaS company that implements solid AEO fundamentals now - schema, llms.txt, content restructure, entity consolidation - is building a moat that compounds over time. AI engines retrain. Citation patterns become self-reinforcing. The company that gets cited first tends to keep getting cited.
The companies I've seen move on this fastest in NZ are the ones with international ambition - where being invisible to an AI tool a US or AU buyer is using to research vendors is a real commercial problem, not a theoretical one.
Where to start
AEO can feel overwhelming because it touches content, technical setup, and brand strategy simultaneously. But the first 20% of the work delivers about 80% of the early impact. Here's the priority order.
- Audit your existing schema: use Google's Rich Results Test and schema.org validator. Most sites have gaps or errors that have been sitting there for years.
- Write and publish llms.txt: plain text, 200–400 words, sitting at
yourdomain.com/llms.txt. Describe what you do, who for, and which pages matter most. - Rewrite your homepage and product pages for direct answers: the first paragraph of every key page should answer "what is this and who is it for" without hedging.
- Consolidate entity language: audit how your company is described across your own site, LinkedIn, Crunchbase, G2, and any media mentions. Standardise the language.
- Add FAQ schema to high-value pages: identify the five questions your buyers actually ask and build FAQ-marked content around each one.
That's a 4–6 week project for a focused operator. It's not glamorous work. There's no launch moment. But six months after doing it, the compounding effect shows up in citation frequency, branded search, and eventually pipeline - and it's very hard for a competitor to undo once you've established the pattern.
AEO isn't a replacement for SEO. It's a new layer that most B2B sites haven't built yet.
Schema, llms.txt, direct content structure, and entity consistency are the four levers. Most NZ B2B sites are missing at least three of them. The window to move first is still open - but it's closing faster than most people realise.
