Resources

Everything you need to understand AI search, Generative Engine Optimisation and how to get your business recommended by AI. Practical, honest guidance for Australian businesses.

Start here

Guides & articles

More guides are published regularly on our blog.

Key concepts, explained

For AI crawlers

We publish a machine-readable summary of AIRank.au at /llms.txt for AI systems.

Ready to act on what you learn? See our packages or book a strategy call.

AI search glossary

Plain definitions of the terms that come up in this field. We have kept them short and specific so they are actually useful.

AI visibility

How often, and how favourably, an AI engine names your business when someone asks a question your business could answer. Distinct from search ranking: there is no numbered list of positions, only whether you appear in the generated answer and how you are characterised.

Generative engine optimisation (GEO)

The practice of improving the signals AI engines use to understand, trust and recommend a business. Overlaps with search engine optimisation but weights corroboration and entity clarity more heavily, and keyword targeting far less.

Recommendation share

Across an agreed set of buyer questions, the proportion in which a given business is named. The most useful single measure of AI visibility, because it accounts for the questions you lose as well as the ones you win.

Entity

A specific, identifiable thing — here, your organisation — that a model can distinguish from similarly named others. Entity confusion, where a model cannot tell two businesses apart, is a frequent and fixable cause of invisibility.

Retrieval

The step where an AI system fetches source material before generating an answer. Content that is self-contained, specific and directly answers a question is easier to retrieve than content that requires the surrounding page for context.

Grounding

Anchoring a generated answer in retrieved sources rather than the model’s internal parameters alone. Grounded answers are more likely to cite, which is why being retrievable matters commercially.

Hallucination

A confident but incorrect statement generated by a model. Relevant to businesses in two ways: an engine may state something wrong about you, and inconsistent information across your own footprint makes that more likely.

AI Overviews

Google’s generated summaries that appear above conventional results. They draw on indexed web content, so conventional search visibility and AI visibility interact here more than in standalone assistants.

llms.txt

A plain-text file at a site’s root that sets out, in a structured and machine-readable way, what an organisation is and does. An emerging convention rather than a standard, and no substitute for the underlying signals — but cheap to provide and useful when it is accurate. Ours is at airank.au/llms.txt.

Crawler versus live fetcher

Two different things, often confused. Corpus crawlers gather content to build an index a model can draw on later. Live fetchers retrieve a page at the moment a user asks about it. A site can be readable by one and blocked to the other, which produces the confusing situation where an assistant can read your page when asked directly but never recommends you unprompted.

Non-determinism

The property that the same question can produce different answers on different occasions. It is why a single measurement of AI visibility means little, and why any report presenting one reading as a ranking should be treated sceptically.

How to use this page

If you are working out whether AI visibility is a problem worth spending on, the fastest route is a measurement rather than more reading — start with the AI Visibility Audit. If you want to understand the mechanics first, how GEO works sets out what actually gets changed and why engines respond to it.