Google AI Overviews — the AI-written summary that now sits above the blue links for many searches — reduce the number of people who click through to websites. That part isn't in dispute, and the data is consistent enough that a business whose traffic leans on informational blog content should expect that traffic to soften. What is worth arguing about is the response. For most businesses, panicking about "losing Google" misreads the situation. The work that pays back is shifting from ranking and waiting for the click to becoming the source the summary quotes, while protecting the pages that still earn clicks reliably. This post lays out what the numbers actually say, which of your pages are exposed, and where we'd put the effort.
What an AI Overview actually is
An AI Overview is the block of AI-generated text Google places at the top of a results page, stitched together from multiple sources and usually citing a handful of them with links. It's no longer a fringe experiment. On Google's Q2 2025 earnings call, Sundar Pichai put AI Overviews at 2 billion monthly users, up from 1.5 billion a few months earlier, across more than 200 countries.
Singapore is squarely inside that footprint. Google switched on AI Mode — the fuller conversational version of the same idea — for Search in Singapore on 25 August 2025, in English, running on a custom version of Gemini 2.5. It uses what Google calls a query fan-out: it breaks your question into sub-questions, runs several searches behind the scenes, and assembles one answer. The practical upshot is that a single search now resolves questions that used to take three or four, and the answer often arrives without a click landing on anyone's site.
The click problem is real, and the numbers back it up
The clearest read comes from Pew Research, which tracked 68,879 real Google searches from a panel of US adults in March 2025. When an AI summary appeared, people clicked a traditional search result in 8% of visits. When no summary appeared, they clicked nearly twice as often, in 15% of visits. And the citation links inside the summary barely absorbed the difference: users clicked one of those in just 1% of visits to a page with a summary. Google publicly disputed Pew's methodology, so treat it as one strong signal rather than gospel.
It isn't the only signal. A randomised experiment by researchers at the Indian School of Business and Carnegie Mellon, run on 1,065 desktop Chrome users and posted in 2026, found that showing an AI Overview cut outbound clicks to websites by 39.8% and pushed up zero-click searches by 34.5%, while paid ad clicks held flat. The same study checked whether the clicks that vanished were just low-quality traffic Google was filtering out, and found no meaningful difference. The clicks a business loses to an Overview appear to be ordinary clicks, not junk.
Two different methods, same direction. If a chunk of your traffic comes from queries Google can now answer in a sentence or two, some of that traffic is already gone or going.
Which of your pages are actually at risk
Not evenly. AI Overviews are hungriest for questions that have a clean, factual answer, and least useful for searches where the person needs to compare, transact, or reach a specific business. That maps onto page types fairly cleanly.
| Query type | Example | Exposure to AI Overviews |
|---|---|---|
| Definitional / how-to | "what is composable commerce" | High: the answer fits in a summary |
| Product research | "best pos system for a cafe" | Medium: people still click to verify |
| Comparison | "shopify vs custom store" | Medium: nuance resists summarising |
| Local / near-me | "web design agency singapore" | Lower: people want a specific provider |
| Branded | "loma digital" | Low: the searcher already wants you |
The businesses hit hardest are the ones whose organic traffic leans on top-of-funnel explainer content. If your best-performing blog post answers a definitional question, expect its clicks to drift down even while its ranking stays put. That gap between holding a position and losing the click is the whole story, and it's why ranking reports have started to flatter accounts that are quietly bleeding sessions. We wrote separately about why rankings and AI-search visibility have come apart; this is the mechanism.
What we're seeing in client accounts
We manage this shift for Singapore businesses across a few sectors, and the pattern repeats. On one B2B services client, the informational articles that used to be the top of the funnel held their Google positions through the first half of the year but shed a visible share of their clicks — the classic AI Overview signature. What changed for the better was harder to see in a traffic chart: the same brand started turning up inside AI answers for the questions those articles addressed, which meant the content was still doing its job, just no longer being paid for it in clicks. Reading that account by sessions alone would have looked like decline. Reading it by presence in the answer looked like progress.
That's the mental adjustment. The value of a well-structured, genuinely useful page is increasingly banked as a citation and a brand impression inside an AI answer, not only as a session in your analytics.
What to actually do about it
Three moves, in order of how much they pay back.
First, make your important pages easy for an AI to quote. That means answering the core question early and plainly, using clear headings a person would actually search, and adding the structured data that tells engines what a page is. This is the same work that earns citations in ChatGPT and Perplexity, not a separate Google-only trick; our guide to getting cited in AI answers walks through the specific changes. The goal is to be the source the summary is built from, because that citation is the new front door.
Second, protect and invest in the pages that still earn clicks. Comparison content, product and service pages, local landing pages, and anything transactional are far less likely to be summarised away. Shifting some content effort from "explain a concept" toward "help someone decide and act" is a rational hedge, not a retreat.
Third, change what you measure. If you judge content by raw sessions, AI Overviews will read as pure loss and you'll cut the wrong things. Track impressions and average position alongside clicks, watch which queries trigger an Overview for you, and check whether your brand is actually named in AI answers. We go deeper on this in our note on measuring GEO performance, and it's the backbone of how our generative engine optimisation work is scoped.
Where honesty is required
None of this comes with a guarantee, and any agency that offers one is selling you something. Google doesn't publish the criteria for what gets pulled into an Overview or which sources it cites, and those choices shift as the models change — Gemini updates have already reshuffled results more than once. Measurement is genuinely immature: Search Console doesn't yet cleanly separate AI Overview impressions, and attributing a sale back to an AI citation is closer to informed inference than hard tracking. The realistic promise is a better chance of being the cited source and a more resilient traffic mix, not a restored click count. For some very transactional businesses the honest answer is that AI Overviews barely touch them and the sensible response is to do nothing dramatic.
The businesses that will do well here are the ones that stop treating the click as the only thing worth winning. The click was always a proxy for attention and trust. Those are still available — they've just moved up the page, into the answer itself.
Frequently asked questions
Will AI Overviews kill my website traffic? Not all of it, and not evenly. Traffic from definitional and how-to queries is most exposed; transactional, comparison, local, and branded searches hold up far better. If your traffic depends heavily on explainer content, plan for softer clicks there and rebalance toward pages that help people decide and act.
Can I stop Google from showing an AI Overview for my content? Not selectively. You can block AI training and some AI features at a broad level, but doing so tends to cost you visibility rather than protect traffic, because you're opting out of where searchers increasingly look. For most businesses the better play is to be cited well rather than excluded.
Is optimising for AI Overviews different from normal SEO? It builds on the same foundation. AI answers lean on sources that already rank and demonstrate authority, so crawlable, well-structured, credible pages still matter. The additions are answering questions directly, structuring content so it's easy to extract, and measuring citations and impressions rather than clicks alone.
How do I know if AI Overviews are affecting me specifically? Look for pages holding their Google ranking while losing clicks — that split is the tell. Check which of your queries now trigger an Overview, and search your own key questions to see whether Google's answer names you or a competitor.
