The Australian AI Visibility Index

The pattern runs through the whole study, and it is simple: if you sell a service rather than a product, AI is answering questions about your prices without ever mentioning you.

None of this stopped the engines answering. They simply answered from somewhere else, pulling prices out of directories, comparison sites, industry bodies and government pages. So when a buyer asks what your kind of work costs, they get a number back. It just isn’t coming from you.

Two numbers matter in this table. Named is how often the AI mentioned one of an industry’s businesses when somebody asked about that industry. Linked is how often the answer included a link to the business’s own website. Being named puts you in the conversation. Being linked makes you the source.

Read the table from top to bottom and the line through it is hard to miss, though it probably isn’t the one you would expect. Size doesn’t explain who gets seen. Neither does age, and neither does revenue. What separates the visible from the invisible is whether you sell a product or a service. Ask about supermarkets or banks and the AI names real brands about 89% of the time. Ask about builders, lawyers or physios and that falls to 51%. And on the questions that actually decide who gets hired, service businesses barely show up at all.

Your industry, in four numbers

AI can talk about you without ever linking to you.

Across the Index, four in ten times an AI mentioned a business, the answer carried no link to that business’s own website. In those answers the AI described the company entirely from what other people have written about it: reviews, directory listings, forum threads, old news stories.

That gap matters more than it first appears. If the machines are describing you from third-party sources, then third parties are writing your sales pitch for you, and nobody is showing you the draft before it goes out.

Sometimes the AI has no idea who you are.

Not everything we found was absence. Some of it was mistaken identity. Asked “Is Up any good? What do reviews say?”, one engine went away and reviewed the Pixar film. Asked about Rebel, another worked through a camera, a movie and an app before it got anywhere near the sports retailer. And one business in the study has a name so generic that across all fifteen of its answers, it was never once identified correctly. An engine decided the name was “an industry term” and recommended other companies instead.

This one is worth sitting with for a moment, because it is not a ranking problem. If an engine can’t connect your name to your business, you are not ranking badly. You simply do not exist in the answer at all. Somewhere along the way, having a distinctive name stopped being a branding preference and quietly became a technical requirement.

Here is the short version. The full methodology is published as its own page: how every business was picked, how the questions were written, how the answers were captured and scored, and the complete register of businesses. That is deliberate: a study that hides its working is asking you to take its numbers on faith, and these numbers are far too pointed for that.

1. We measured 100 Australian businesses across 10industries. The consumer categories were picked by market position from public rankings. The service categories were picked by who actually ranks on Google for the category’s own search terms, with directories and aggregators taken out. Three national brands failed that ranking test and we kept them anyway, and said so, because a brand missing from its own category is a finding rather than a sampling error.
2. We wrote 473distinct questions of five kinds: who’s the best, what does it cost, this one or that one, is this brand any good, and how do I fix this. Every one of them is a question a real buyer would type, not a keyword dressed up as a sentence.
3. We put every question to all three engines under identical conditions and captured every answer word for word, along with its date, model and sources. ChatGPT ran on gpt-5.5-2026-04-23, Perplexity on sonar, and Google AI Overviews via DataForSEO, geo-coded AU.
4. Every answer was then scored the same four ways: did it name the business, did it link to the business’s own site, did it get its facts right, and did it recommend somebody else instead. Of 1,500 business and question pairs, 1,459 came back with an answer we could score and 41 did not.

Nothing in this Index suggests that AI visibility is a ranking you can go out and buy. What it does tell you comes down to three things.

1. The questions that actually win you work, like what things cost and who is any good, are being answered right now without you anywhere in them.
2. Engines build their answers out of pages that already exist. Directories are winning because they publish exactly what buyers ask for, prices most of all, and most service businesses simply don’t.
3. Each engine rewards something different, and all three of them can be measured. This Index is that measurement run at industry scale, and we run the same one for individual businesses.