Which AI Engines Name Australian Clinics

Yes, AI engines name Australian clinics, and they mostly quote clinic websites. On 12 September 2026 we put 12 Australian patient questions to four AI engines. Clinic and practice websites took 60% to 72% of all citations. Directories such as HealthEngine took only 6% to 21%. The catch is overlap. Of the 167 clinic websites cited, 74.3% were cited by exactly one engine, and only six were cited by all four. Google showed no AI Overview at all for 3 of the 12 questions, including the most commercial one.

Which AI engines actually name Australian clinics?

All four engines name clinics, and they do it far more often than most practice owners expect. What differs is how often, and how widely they spread the names around.

On 12 September 2026 we put the same 12 questions to four AI answer tools. The questions are the kind your patients type. They came in four groups of three. One group asked for a provider in a named suburb. One gave a symptom and a city. One named a service. The last three were general questions with no place in them, as a control.

We saved every answer. We cut every citation down to its web domain and sorted it. Here is how often each engine cited a provider's own website in its answer.

Perplexity cited a clinic in every answer. Google AI Overviews managed 8 of 12. Part of the reason is that it refused to give an overview at all three times. That is a finding of its own, and we cover it further down.

Across all four engines, 167 distinct Australian clinic and practice websites were cited. So being named by an AI engine is quite normal for a clinic. It is far from a rare prize.

The engines differ most in how widely they spread their citations. Perplexity cited 125 different provider sites across its 12 answers. ChatGPT cited 31.

That gap matters for your business. On Perplexity your clinic is one in a crowd of 125. On ChatGPT you are one of 31. So the same work to get seen does not buy you the same thing on each.

One engine is missing from this table, and we want to be upfront about it. We asked Gemini all 12 questions. It gave an API credit error every time, so we recorded it as not assessed. That does not mean it scored zero. We simply could not ask it, and that is very different from Gemini not seeing you.

Do AI engines send people to directories or to clinic websites?

They send people to clinic websites, by a wide margin. This result goes against the advice most clinics are given.

The usual advice is that AI search runs on directories. So you are told to pour your effort into HealthEngine, HotDoc, Cleanbill and the health fund finders. We sorted every cited domain to test that advice.

Google AI Overviews cited a directory in 6.2% of cases. Of the four, it leaned on directories the least.

Perplexity leaned on directories the most, at 21.3%. Even so, it cited clinic websites three times more often than directories.

Here is what this means for you, stated carefully. Your directory listings may still be worth keeping for booking, for reviews and for patients who search that way. But they were not what these four engines mostly quoted. The engines mostly quoted the clinic's own page.

If one engine cites your clinic, do the others?

Usually not. This finding is the hardest one to hear.

Of the 167 clinic websites cited across the four engines, here is how many engines cited each one.

Exactly one engine cited 124 of 167 sites, or 74.3%. Only six sites were cited by all four.

So showing up in AI search is not one single thing your clinic has or lacks. On this evidence there are four separate races. If you win one, it tells you almost nothing about the other three.

There is a real pull the other way too, and leaving it out would be unfair to the data. Look at one question at a time, and the engines agree far more than the 74.3% suggests. There were 11 questions where at least two engines cited any clinic at all. In 11 of the 11 questions, at least two of them cited the same clinic.

Both things are true at once, and that mix is the real finding. Each question has a small shared core of clinics that most engines land on. Around that core sits a long tail of clinics that only one engine names. If your clinic is in the core, you are seen everywhere. If you are in the tail, you are seen in one place, and you have no way of knowing it.

Why did Google refuse to answer the most commercial question?

Google is most careful on the questions closest to a booking, and this was the most commercial question.

Three of the 12 questions got no AI Overview at all. Google did not even show a thin one.

Three cases do not prove a pattern, and we will not claim they do. But they hint at one, and they fit what Google says about how it handles health.

"Best physiotherapist in a named suburb" asks Google to rank named health workers against each other. "Where can I get a skin cancer check" asks it to point someone toward checking for a possible cancer. Google treats both as high-stakes. On both, it showed nothing and let the normal search results do the work.

For your clinic, that is good news. On the searches closest to a booking, the classic blue link was still the whole answer. So your Google ranking still matters.

What happens when an engine gets the country wrong?

It sends your patient to the wrong side of the world, and it sounds sure of itself while doing it.

Seven citations in this run pointed to groups outside Australia. All seven came from one engine, Perplexity, and they sat in two answers.

The Newcastle case is the clearest. There are two well-known Newcastles. The engine picked the English one, even though the question said New South Wales.

This is a quiet reason to put your suburb, your state and your country in plain text on your own pages. Hidden code that people never see is not enough on its own. Write it into the sentences your patients read.

If an engine cannot tell which Newcastle you are in, it will not guess in your favour.

How was this measured, and what did the machine get wrong?

We measured it in the open. The machine got about a fifth of it wrong before a person fixed it.

The questions. We wrote 12 questions before we collected any data. They came in four groups of three. Three asked for a provider in a named suburb. Three gave a symptom and a city. Three named a service and a city. Three were general questions with no place in them, as a control.

The engines. We used ChatGPT through the OpenAI Responses API, with web search and an Australian user location. We used Claude Sonnet 4.5 with a web plugin, and Perplexity Sonar. We also captured live Google AI Overviews from the real Australian results page. Every engine got the same 12 prompts and the same neutral instruction. We saved every raw answer.

The sorting. We cut every cited web address down to its domain. Then we put each one in one of four groups. These were a provider's own website, a directory or comparison site or insurer, a peak body or government or research source, or could not be checked.

The mistake, and this is the part worth reading. The first pass was automatic. It counted any domain not on a known directory or government list as a clinic website. That counted far too many. A person then opened all 212 flagged domains. They read each site's own homepage title and description, and moved the ones that were not a clinic.

The machine got 45 of 212 wrong, an error rate of 21.2%. YouTube was in there. So was PubMed. So were three dictionaries, a university and a consumer group. There were also 23 directories and comparison sites the keyword list had never heard of.

Every number in this post is the hand-checked version. We publish the error rate for a reason. Any tool that sells you an automatic "AI visibility score" is running the pass we ran first. It is very unlikely to tell you how often it is wrong.

Nine domains would not load at all. We counted them as unchecked rather than guess.

We have chosen not to publish the names of the Australian clinics that were cited. A list of named practices reads like a recommendation. A record of what an AI said this week is not a recommendation.

What does this change for a clinic?

It changes three things for you, and the first one is the cheapest.

Check it yourself, on each engine, and write down what you find. Ask each engine the three or four questions your patients really ask, in their own words. Note which clinics come back and what gets cited. Do it on all four, because 74.3% of cited clinics showed up on only one. One check on one engine tells you close to nothing.

Work on your own pages before the directories. Between 60% and 72% of what these engines cited was a provider's own website. Say you must choose between improving a service page and topping up a directory profile. This data points you to the service page.

Say where you are, in words. The country mix-ups in this run were not subtle. Put your suburb, city, state and country in the visible sentences on the page. Hidden code alone is not enough.

One word of caution on the method. This is 12 questions on one day, and engines change weekly. The useful part is the method you can repeat. The exact percentages matter less. Run it again next quarter and the numbers will move. The point is that you can run it again at all. You cannot say that about most claims made about AI search.

Method and sources. 12 prompts × 4 engines, run 12 September 2026. ChatGPT via the OpenAI Responses API (gpt-5-mini, web_search tool, user location AU); Claude Sonnet 4.5 and Perplexity Sonar via OpenRouter; Google AI Overviews captured live from the Australian result page. Gemini returned an API credit error on all 12 prompts and is recorded as not assessed. We sorted 427 citations in total. We checked 212 domains by hand against their own homepage titles, and moved 45 of them. We kept the raw answers, the domain evidence and the sorting so you can repeat the run.

Related reading:

• How to rank in AI search
• When Google answers your patient's question for you
• What patients read on a clinic's About page before they book
• SEO and GEO fix for clinic websites
• View all services →