Has AI search changed how Singapore buys, and the Singapore figure to stop looking for

The supply side has changed and we can prove it. The demand side has no Singapore number, and anybody quoting one is quoting an estimate somebody built from a global survey. So the useful answer to this question is in two halves: what is measurable about what the engines now say to a Singapore buyer, and how to get your own figure for whether your own buyers are using them.

Start with the figure to stop looking for. There is no published, reproducible statistic for the share of Singapore buyers who consult an AI assistant before contacting a supplier. We have not run that study, nobody we can check has run it for this market, and a page that prints a percentage for it is printing a guess with a decimal point on it. What follows is what we did measure.

The supply side, measured on 22 October 2026

We put 33 batches of real buying questions to ChatGPT and Gemini with Singapore written into the text of each one. The market has to go in the prompt because the search interface we use accepts country, language and location parameters, answers HTTP 200 to all of them, and ignores every one. The snapshot taken that evening holds 64 blind readings, 32 on each engine, with the brand batch excluded.

Outcome	ChatGPT, 32 readings	Gemini, 32 readings	Both, 64 readings
Named a business that was not us	19	19	38
Named no business at all	13	13	26
Named us	0	0	0
Cited our domain	0	0	0

So the answer a Singapore buyer gets today is a real answer with real suppliers in it 38 times in 64, and an answer with nobody in it 26 times in 64. The engine split is worth keeping separate and never averaging, and in this market the separate figures turn out to be the same. Each engine named a business in 19 of its 32 readings and named nobody in 13.

What the engines named when they named anybody

Six Singapore agencies carried almost all of it. First Page in 17 of the 64 readings, MediaOne in 12, Impossible Marketing in 11, OOm in 10, Clickr in 2 and Verz in 1. On the tool side, Semrush in 13, HubSpot in 12, Ahrefs in 9, Profound in 5, then Peec and Otterly in 3 each.

That concentration is the evidence that behaviour has changed on the supply side. A Singapore buyer who asks for a shortlist is handed roughly the same four firms, which is a narrower answer than the first page of a search result and a far narrower one than a directory. Being inside or outside that set of four is now a commercial fact about a Singapore agency, and none of those four bought their way in.

What changed on the supply side that a seller can act on

Three things are different from a search result, and each one changes a decision.

• The shortlist is shorter. A first page of search results carries ten organic links and several ads. These answers carried four Singapore agency names doing most of the work. A buyer who used to compare eight suppliers is being handed four.
• The sources are not the brands. The hosts behind these answers were small Singapore specialist sites and government pages, not the named agencies' own homepages. So the thing that got an agency named was often somebody else's page about the category, which means publishing about your own category is how you enter the set.
• An answer can contain nobody. A search result always lists somebody. An assistant can decline, and in this market it declined 36 times in 64. That failure mode does not exist in search and it is the one worth watching.

The method that gives you your own demand number

You do not need a market study. You need two questions on your own enquiry form and one quarter of patience.

• Ask every enquirer, in plain words, where they first heard about you. Free text, not a dropdown, because a dropdown gives you the options you thought of.
• Ask whether they checked anything before contacting you, and what. This is the question that surfaces an assistant, because an assistant visit leaves no referrer and your analytics cannot see it.
• Count branded search separately. A buyer who was told your name by an assistant and then searched for it arrives in analytics as organic brand traffic. Rising branded search with flat non branded is the signature to watch for.

A quarter of that gives you a defensible Singapore figure for your own business, which is worth more than any national percentage. Our page on what a good mention rate is, without inventing an industry benchmark covers how to read the supply side half once you have it, and the same question measured for India shows what the answer looks like in a market where the readings came out differently.

Questions about AI search and Singapore buying behaviour

Is there a reliable figure for how many Singapore buyers use ChatGPT before buying

Not one we can check, and that is the honest state of it. The measurable half is what the engines say when asked, and that half is on this page with its date and its denominator.

Does this cover Google AI Overviews

No. This run is ChatGPT and Gemini. Perplexity, Claude and AI Overviews were not in it. Our product watches five surfaces and this dataset covers two, and the two never get merged into one sentence.

Why is the Gemini result different from the ChatGPT one

They are different retrieval systems and they behave differently, which is why we report them separately. In this Singapore set each engine named a business in 19 of its 32 readings. Averaging those two would describe a market that does not exist.

How often should a Singapore business retake this reading

Monthly is enough to see movement and cheap enough to sustain. The scorecard behind this page was still being written on the evening of 22 October 2026, which is a fair illustration of how quickly a number in this category stops being current.

Does any of this apply to a purely local Singapore business

It applies more, not less. A city state has no geographic spread to hide behind, so the set of businesses an assistant can name for a Singapore category is small, and being absent from it is more visible here than in a large country.

The number that is not in anybody's analytics

A Singapore business can see its search traffic, its ad costs and its enquiry volume. What it cannot see from any of those is the conversation where a buyer asked a machine for a shortlist and your name was or was not in it. Those 26 supplier free answers out of 64 are the clearest sign that the conversation is still being settled, and the 0 beside our own name in the same table is the clearest sign that we are measuring it honestly rather than selling a flattering number. AI Knows Us puts a fixed list of your buyers' questions to ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews on a schedule and shows the sources behind each answer. For a Singapore business the first useful output is not a score, it is finding out which of your own questions still have nobody in the answer.