Google AI Overviews in Singapore: What They Are and What Actually Changed
Somewhere between typing a question into Google and seeing ten blue links, a paragraph appeared. It answers the question directly, cites three or four sources in small text, and it is the first thing on the page. For a Singapore business that has spent years earning position three for a valuable query, the honest question is simple: does that paragraph cost me money, and how much?
The answer is more interesting than either of the two positions you usually hear. AI Overviews have not killed search traffic, and Google’s own revenue makes that claim untenable. But they have measurably changed click behaviour, and they have quietly rewritten which pages get shown as sources — in a way that has nothing to do with your rank.
This guide is the explainer: what the surface is, how often it appears, what the strongest available data actually shows, and what it means specifically in a market where Google holds more than nine in ten searches. It deliberately stops short of tactics. If you want the practical playbook for appearing in these answers, that is how to show up in AI Overviews.
What an AI Overview is, precisely
An AI Overview is a generated summary that appears above the standard results for some queries, with links to supporting pages. AI Mode is the separate, fuller conversational experience, where the entire result is a generated answer you can follow up on. Both are treated as one family in Google’s own reporting.
Two mechanical points matter more than the definitions.
First, eligibility runs through normal search. Google’s documentation on AI features states that to be shown as a supporting link, “a page must be indexed and eligible to be shown in Google Search with a snippet”. There is no separate index and no separate submission process. If a page is blocked, noindexed, or set to nosnippet, it is out of AI answers by the same action.
Second, the answer is assembled from more than the page you can see. Google describes AI features as displaying “a wider and more diverse set of helpful links associated with the response than with a classic web search”. In practice the system decomposes a question into related sub-questions and draws sources across that wider set — which is why the source list often contains pages you would never find by searching the original query yourself.
How often do they appear?
This is where published figures diverge sharply, and it is worth being honest about that rather than quoting the number that suits an argument.
| Source | Reported share of queries with an AI Overview |
|---|---|
| Semrush | Approximately 48% of queries |
| Similarweb | Approximately 43% of queries |
| Range across 2026 industry analyses | Roughly 43% to 64%, depending on keyword sample and device |
A twenty-point spread between reputable vendors is not sloppiness; it reflects that each is measuring a different keyword universe. A tool tracking mostly informational, question-shaped keywords will report high coverage. One tracking transactional and branded terms will report low coverage. Neither is your website.
The consistent finding underneath the disagreement is that trigger rates skew heavily by intent. Informational and question-shaped queries trigger AI Overviews far more often than transactional or branded ones. For a Singapore business, that has a direct practical reading: your “what is”, “how much does”, “how to choose” content is squarely in the affected zone, while “book a demo”, “[brand] price list” and “[brand] contact” queries are largely untouched. Before assuming a site-wide problem, check which half of your traffic actually sits in the exposed category.
What the click data actually says
The strongest independent evidence on click behaviour comes from the Pew Research Center, which is worth preferring over vendor studies for a specific reason: it measured real browsing behaviour rather than modelled estimates, and it published its sample.
Pew tracked over 900 US adults who consented to a browsing tracker, covering 68,879 unique Google searches during March 2025. The findings:
- When an AI summary appeared, users clicked a traditional search result on 8% of visits. When no summary appeared, they clicked on 15%.
- Browsing sessions ended on 26% of pages with an AI summary, against 16% of pages with only traditional results.
- Users clicked a link inside the AI summary itself on just 1% of those results.
Three caveats belong with those numbers, and most coverage omits them. The panel was US, not Singaporean. The data is from March 2025, and the surface has changed since. And it measures correlation on pages where summaries appeared — and summaries appear disproportionately on informational queries, which historically had lower click rates anyway. The direction of the effect is solid; the exact magnitude for your Singapore site is not something this study can tell you.
The 1% figure is the one with the clearest strategic implication. If citation inside an AI Overview almost never produces a click, then being cited is best understood as a brand-visibility outcome rather than a traffic channel. That is a genuinely different thing to measure, and it is why the reporting question needs its own treatment — see AI brand visibility monitoring.
The finding that should change how you think about rank
The more consequential shift is not about clicks at all. It is about which pages Google chooses as sources.
Ahrefs analysed 863,000 keyword SERPs and 4 million AI Overview URLs in March 2026. Only 38% of pages cited in AI Overviews also ranked in the top 10 for that query. Pages ranking 11 to 100 accounted for 31.2% of citations, and pages beyond position 100 for another 31.0%. In their earlier July 2025 study, the top-10 figure had been 76%.
A drop from 76% to 38% in seven months is not a ranking-algorithm story. It is the consequence of query fan-out: the system breaks a question into sub-queries and selects sources that perform well across that wider cluster, not sources that win the original query.
Read this carefully, because it cuts both ways. If you rank in the top 10, your citation is far less automatic than it was a year ago. If you do not, you are no longer excluded — nearly two-thirds of citations now go to pages outside the top 10. For a mid-sized Singapore business that has never cracked page one against the big directories and international publishers, that is the first genuinely encouraging structural change in search in some years.
The Singapore picture
Two facts frame this market specifically.
Google’s share here is unusually high. StatCounter’s July 2026 data puts Google at 92.46% of Singapore searches, Bing at 3.37%, Yahoo at 0.88% and DuckDuckGo at 0.83%. In markets where Bing holds double digits, “diversify away from Google” is a real strategy. Here it is not. Whatever Google does to its results page happens to more than nine in ten of your searchers, and there is no meaningful second channel to hedge into.
Google’s search business has grown through the change. This matters because the loudest prediction about AI search has already failed. Alphabet’s Q2 2026 results reported total revenue of USD 119.8 billion, up 24%, with Google Search and other revenue at USD 63.3 billion, up 17% year on year — a full year into the AI Overviews and AI Mode era. Google also reported record search usage during the World Cup, and the Gemini app at 950 million monthly active users.
Both things are true at once: individual informational pages lose clicks, and total search volume and monetisation grow. If you have been told that AI Overviews are ending search, the accounts say otherwise. If you have been told nothing has changed, the Pew data says otherwise. The reality is a redistribution, not a collapse — a theme we take further in the future of SEO in an AI world.
Local and multilingual queries
Two Singapore-specific wrinkles are worth flagging. Queries with clear local intent — “near me”, “in Singapore”, specific neighbourhoods — behave differently, because the map pack and local results still occupy the top of the page and the AI answer often sits below or alongside rather than replacing them. Businesses with a physical location are less exposed than pure information publishers, which is one reason local fundamentals remain the highest-return work for many SMEs. That groundwork is covered in our local SEO guide.
Second, Singapore’s multilingual search behaviour means the same commercial intent arrives in several languages and registers. AI answers are generated per query, so a business that has only ever optimised for one phrasing of a question is visible in a narrower slice of the AI surface than it assumes.
What Google says you should do about it
Google’s position is unambiguous and worth quoting because so much of the market sells the opposite. From Google’s documentation on AI features: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” And: “You don’t need to create new machine readable files, AI text files, or markup to appear in these features.” Google adds that “There’s also no special schema.org structured data that you need to add.”
Three practical consequences follow.
- There is no separate AI SEO discipline to buy. Anyone selling “AI Overview optimisation” as a distinct service line, with its own file formats and its own audit checklist, is selling something Google’s own documentation says does not exist. The work is search fundamentals, applied to a surface that reads pages differently. That framing runs through our AI SEO guide.
- Your controls are the ones you already have. Google states that to limit what appears from your pages you “use
nosnippet,data-nosnippet,max-snippet, ornoindexcontrols”. These are blunt: restricting snippets restricts your classic listings too. - Structured data is not the lever. It remains worth having for rich results and clarity — see our schema markup guide — but it is not an AI Overview entry ticket.
How to measure your exposure
Measurement is the weakest part of this whole area, and pretending otherwise is how bad decisions get made. Here is the honest state of the instrumentation.
| Source | What it tells you | What it will not tell you |
|---|---|---|
| Search Console Generative AI performance report (from June 2026) | Impressions in AI Overviews and AI Mode, by page, country and device | No clicks, no CTR, no position, no query data. Rolling out to a subset of site owners only |
| Search Console Performance report | Your total search picture, which already includes AI feature impressions | Cannot separate AI from classic without the report above |
| GA4 traffic reports | Sessions and behaviour of whoever arrives | AI Overview clicks arrive as ordinary Google organic traffic and cannot be isolated |
| Year-on-year impressions vs. clicks | The most useful proxy: impressions flat or rising while clicks fall is the signature | Confounded by seasonality and ranking changes |
The single most important trap: AI Overview impressions and clicks are not a separate channel in GA4. GA4’s AI-related channel captures referrals from assistants such as ChatGPT and Gemini, not Google’s own AI surfaces, which stay in Organic Search. Anyone reporting “our AI traffic” from a GA4 channel report is under-counting Google entirely. The practical reporting approach is set out in the GA4 reports worth checking, and the wider attribution discipline in how to measure marketing ROI.
A simple diagnostic you can run this week
Open Search Console, set a 16-month range, and compare impressions against clicks for your top informational pages. Three patterns and their readings:
- Impressions steady, clicks falling. The classic AI Overview signature. You are still being shown; fewer people need to click.
- Both falling. Not an AI story. This is a ranking or indexing problem, and should be diagnosed as one.
- Impressions rising, clicks flat. Usually healthy — you are gaining reach on queries where the answer is satisfied on the page.
Do this per page group, not site-wide. Averaging your commercial pages together with your blog hides the entire effect.
What genuinely changes for a Singapore business
Stripped of the noise, four shifts are real.
1. The value of pure informational traffic falls. Content whose only job was to answer a factual question and collect an ad impression or a newsletter signup is worth less than it was. Content that answers a question and demonstrates why you are the firm to hire is worth the same or more.
2. Rank stops being the whole game. With nearly two-thirds of citations coming from outside the top 10, a rank report is now an incomplete picture of visibility. It is still the best single predictor of clicks, but it no longer predicts citation.
3. Being named beats being linked. If only 1% of AI summary views produce a link click, the value of appearing is that the reader saw your name attached to a credible answer. That is a brand outcome, measured over months in branded search volume and direct enquiries — not in next week’s sessions.
4. Middle-of-the-page competition opened up. The old top-10 bottleneck was the hardest structural barrier for a smaller Singapore firm competing against directories and global publishers. It has loosened considerably. The practical implications of that are the subject of SEO for AI search, and the vocabulary around it is untangled in GEO vs AEO vs LLM SEO.
What has not changed is the answer to whether search is worth investing in at all. The economics still hold, for the reasons set out in is SEO worth it in Singapore — and Google’s own 17% search revenue growth is a reasonable independent check on that view.
The bottom line
AI Overviews took a share of the clicks that used to follow informational queries, and Pew’s data puts a credible number on it. They did not end search, and Alphabet’s accounts put a credible number on that too. The larger change is quieter and more useful: the source list is no longer drawn mainly from the top 10, which means visibility and rank have come apart for the first time in twenty years.
For a Singapore business, the correct response is neither panic nor a new budget line for a discipline Google says does not exist. It is to work out which half of your traffic is exposed, measure impressions against clicks honestly, and keep building the pages that earn a citation on merit.
You can see how that plays out for real Singapore companies in our case studies. If you want a clear read on your own exposure, our AI SEO service starts with exactly that diagnosis. Talk to us and we will tell you what your data says, not what the headlines say.
Frequently asked questions
Do AI Overviews reduce website traffic?
For queries where they appear, yes, measurably. Pew Research found users clicked a traditional result on 8% of visits when an AI summary was present, against 15% when it was not, across 68,879 searches by over 900 US adults in March 2025. The effect is concentrated in informational queries; transactional and branded queries are far less affected. The study is US-based, so treat the direction as reliable and the exact magnitude as something to verify in your own Search Console data.
How do I get my Singapore business into AI Overviews?
There is no separate submission or optimisation process. Google’s documentation states there are “no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary”, and that no AI text files or special schema are needed. Eligibility requires only that the page is indexed and can be shown with a snippet. Beyond that, the work is ordinary search quality, covered in our guide to showing up in AI Overviews.
Can I stop my content appearing in AI Overviews?
Partly, using existing controls. Google names nosnippet, data-nosnippet, max-snippet and noindex as the mechanisms for limiting what is shown from your pages. The trade-off is severe: these controls also restrict your normal search snippets, so you generally lose more classic visibility than you protect.
Does ranking number one still matter if AI Overviews are on top?
Yes, for clicks — position one still earns far more than any other slot. But rank no longer determines citation. Ahrefs found only 38% of AI Overview citations come from pages ranking in the top 10, down from 76% seven months earlier, with roughly a third coming from pages beyond position 100. Rank and AI visibility are now two different measurements.
How do I track AI Overview performance in Google Analytics?
You largely cannot. Clicks from AI Overviews arrive as normal Google organic traffic in GA4 and are indistinguishable from classic organic clicks. GA4’s AI-assistant channel covers referrals from chat assistants, not Google’s own AI surfaces. The nearest first-party view is the Search Console Generative AI performance report, which provides impressions only — no clicks, CTR, position or query data — and is still rolling out to a subset of sites.
Are AI Overviews a bigger deal in Singapore than elsewhere?
The exposure is higher because the dependence is higher. StatCounter’s July 2026 data puts Google at 92.46% of Singapore searches against Bing’s 3.37%, so there is no meaningful alternative channel to shift budget into. That said, businesses with a physical location are somewhat insulated, because local intent queries still surface map and local results prominently.
Related guide
This post covers what the surface is and what the click data shows. For the measurement side — how to see any of this in your own analytics, and why GA4 will never separate AI Overview clicks from ordinary organic ones — see how to track AI search traffic in Singapore.
