Sign inStart free

Is there Australian data on AI assistant use for supplier choice, and the supply side reading that does exist

No published Australian demand figure exists, so here is the supply side reading with its engine split, and the two questions that give you your own number.

Published by AI Knows Us · Updated 22 October 2026

There is no published, reproducible Australian figure for the share of buyers who consult an AI assistant before choosing a supplier. We have not run that study, and we have found none for this market that names its sample, its method and its date. Anybody quoting a percentage is quoting an estimate built from a global survey and applied to Australia, which is a guess with a decimal point.

What does exist is a supply side measurement, which answers a different and more actionable question: what the assistants say when an Australian buyer asks. That reading is on this page with its denominator, its engine split and its date, and the second half sets out how to produce your own demand side number in a quarter.

The supply side reading, with its method

On 22 October 2026 we put 33 batches of buying questions to ChatGPT and Gemini with Australia written into the text of every question. The market has to go in the prompt because the search interface we use accepts a country parameter, a language parameter and a location parameter, returns HTTP 200 for each, and ignores all three. The snapshot taken that evening holds 64 blind readings, 32 on each engine, with the brand batch excluded because a question naming us is a ceiling rather than a result.

OutcomeChatGPT, 32Gemini, 32Both, 64
Named a business that was not us162036
Named no business at all161228
Named us000
Cited our domain000

Report the two engines separately and never average them. Gemini named a business in 20 of its 32 readings and ChatGPT in 16 of its 32, and a mean of those describes a market that does not exist.

What it named when it named anybody

Six Australian agencies carried almost all of it. Online Marketing Gurus in 12 of the 64 readings, Rocket Agency in 8, StudioHawk in 8, Prosperity Media in 8, Sparro in 3 and King Kong in 3. On the tool side, Semrush in 14, Ahrefs in 12, HubSpot in 11, Otterly in 7, Peec in 4 and BrightLocal in 3.

That concentration is itself a usage finding. A buyer asking a machine for an Australian shortlist is handed four to six names, which is narrower than a page of search results and much narrower than a directory. Whether many Australian buyers do this or only a few, the ones who do are being funnelled hard.

The Australian source pattern, which no survey would show you

The engines returned 1,010 distinct hosts across those 64 readings, 435 of them on Australian domains, and 66 distinct Australian government hosts appeared 139 times across 25 of the 64 readings. business.gov.au in 15, accc.gov.au in 13, acma.gov.au in 8, oaic.gov.au in 6, legislation.gov.au in 5, business.qld.gov.au in 5.

That is a structural fact about how Australian commercial questions get answered and it is more useful to a seller than any adoption percentage. The buyer is being handed the state's view of your category alongside the supplier names, which means a supplier page that answers to the rules is a stronger candidate than one that avoids them.

The two questions that give you your own demand number

  1. Where did you first hear about us. Free text, not a dropdown, because a dropdown returns the options you thought of.
  2. Did you check anything before contacting us, and what. This is the question that surfaces an assistant, because an assistant visit leaves no referrer and your analytics cannot see it at all.

Then watch one derived signal. Branded search rising while non branded stays flat is the signature of somebody being told your name elsewhere and then looking you up. It is free in Search Console and it is the closest thing to an attribution trail this channel leaves.

What an Australian business can honestly say to a board

Three sentences and no more. First, that there is no reliable national figure for how many Australian buyers consult an assistant, and anybody offering one should be asked for its sample size and its date. Second, that what the assistants answer is measurable and has been measured, with the denominator, the engine split and the date attached, and that the answer currently contains no supplier at all in 28 of 64 readings while naming somebody, usually a local agency, in the other 36. Third, that the business own exposure is knowable from its own enquiry form inside a quarter at no cost. A board paper built on those three is defensible under questioning. One built on a borrowed adoption percentage comes apart the moment somebody asks where the number came from, and that question will be asked.

Why no credible Australian figure exists yet

Three reasons, and they also tell you what to distrust. The behaviour is new enough that any survey measures a moving target. Self reported tool use is unreliable, because people under report asking a machine. And Australia's cell in a global survey is usually too thin to report, so a national figure gets derived rather than measured. A derived figure presented as a measurement is the thing to refuse.

Where ours is weak, we say so. These 64 readings are two engines on one date. Perplexity, Claude and Google AI Overviews were not in this run. There is no per industry rate in it, because the questions spanned sectors rather than sampling one. And the scorecard behind this page was still being written on the evening of 22 October 2026, which is why these figures come from a frozen snapshot rather than a running total.

Questions about Australian AI usage data

Why will you not give a percentage

Because we would have to invent it. Every figure on this page names the document or dataset it came from and the day it was read, and the demand side number does not exist in any source we can check.

What is the most useful number for my own business

The share of your own buying questions that currently return an answer with no supplier in it. Specific, checkable, and it tells you what to publish, which no national percentage would.

How does this compare with other markets

It has to be measured separately per market, which is why we ran seven countries rather than one. Our page on the same question measured for India gives a different answer from a different reading, and the two are not interchangeable.

Does the free scan give me a usage figure

It gives the supply side half, which is what the engines say about your category, and that is the half that tells you what to do. The demand side half comes off your own enquiry form.

How often should this be remeasured

Monthly for the supply side, quarterly for the demand side. Our own scorecard moved inside a single evening, which is a fair illustration of the shelf life.

The structural fact that beats an adoption percentage

If you only take one number from this page, take 139. That is how many times 66 distinct Australian government hosts appeared across 25 of our 64 blind readings on 22 October 2026, and it is more useful to a seller than any national adoption figure would be. It means an Australian commercial question gets answered partly by the state: business.gov.au in 15 readings, the consumer regulator in 13, the communications authority in 8, the privacy regulator in 6, the statute register in 5, state portals from seven jurisdictions. A supplier page that answers to that framing is a stronger candidate than one that avoids it, and that is a publishing instruction you can act on tomorrow. An adoption percentage, even a real one, would change nothing you do. AI Knows Us runs the same measurement for a business across ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews and reports named and cited separately because they fail for different reasons. What it will not do is hand you a national usage statistic, because nobody has one that names its sample, its method and its date, and we would rather publish the method than the guess.

Read next

Simon-Kucher pricing strategy, what the firm publishes about its own method and what a Singapore buyer should ask before briefing itThe firm publishes its approach, its scale and four named service lines, and publishes no fee, so the useful work before any call is reading what it claims and deciding which decision you are buying.International pricing strategies for a Singapore company, and the corridor that stops one market undercutting anotherA converted home price is the one answer that is almost always wrong, so set a price per market and keep the gaps narrower than the cost of moving a unit between them.Premium pricing strategy for a Singapore business, and the fence that makes the higher number holdCharging permanently more than the credible alternatives works when the buyer can name what the extra money buys, and the published Singapore writing ladder shows exactly how that is built.Red Antler publishes no price, so here is what its own site does publish and how to cost the same scope from SingaporeRead on 29 October 2026, three pages of redantler.com carry a complete deliverable list, three business stages and not one figure about what any of it costs.Hiring a full service content agency, the five stages a retainer has to cover and the two handoffs where the money leaksMost content retainers buy production and quietly leave strategy and distribution with you, so price the stages rather than the word count.Hiring a branding agency for a Singapore financial services firm, and the three registers to search before you fall in love with a nameIdentity work for a licensed firm starts with a clearance search rather than a mood board, and two of the three registers are free to read in a browser today.

Or browse the whole archive.

See what AI says about you.

The first scan is free, it needs no account, and it asks the questions your buyers ask.

Free. No card. We ask 5 real buyer questions on 2 AI apps.

Already sure you want this? Start free. No card needed, and the free plan stays free. Pay only when you want the work done. Or talk to a person if you have a question first. Or read how we fix what it finds.