Google no longer just points people to answers. It writes the answer, on the page, and a lot of the time nobody clicks anything after reading it. If your traffic charts have looked strange since late 2025, this is most of the story.
I have spent the last year watching client analytics dashboards tell the same story from different industries: impressions holding steady or even rising, clicks sliding down. That gap has a name now, and it sits at the top of nearly every search results page.
Google AI Overviews are AI-written summaries that appear automatically above the blue links, and AI Mode is a separate conversational search experience you open on purpose, where you can ask follow-up questions and refine your query without ever leaving Google. Both are powered by Google’s Gemini models, and together they are reshaping how people find businesses like yours. Here is what is actually happening and, more usefully, what to do about it.
What are Google AI Mode and AI Overviews?
Let me separate the two, because the names get used interchangeably and they are not the same thing.
AI Overviews are the summary box you did not ask for. You type a query, and before you reach a single organic result, Google shows a few paragraphs or bullet points of AI-generated answer, stitched together from multiple sources and topped with a handful of citation links. You did nothing to trigger it. Google decided the query deserved a summary and produced one. As of early 2026, AI Overviews appear on roughly 48% of all Google queries, up sharply from around 34% at the end of 2025.
AI Mode is the summary box you did ask for. It is a distinct tab in Google Search, a full conversational interface where you can ask a messy, multi-part question, get a synthesised answer, then follow up in plain language. Instead of running three separate searches, you hold one conversation. AI Mode passed one billion monthly active users roughly a year after launch, which tells you the behaviour is not niche.
The technical distinction matters for strategy. AI Overviews summarise a single query from Gemini’s indexed crawl. AI Mode uses a more capable reasoning model, runs what Google calls query fan-out (it silently issues many related sub-queries behind your one question), and pulls live signals like stock prices and Shopping Graph data. AI Mode is doing more work, from more sources, per question.
What is the difference between AI Mode and AI Overviews?
The short version: AI Overviews replace the blue links with a synthesised answer above them; AI Mode replaces the whole search session with a conversation. Here is the practical breakdown.
| Classic search | AI Overviews | AI Mode | |
|---|---|---|---|
| How it starts | User types a query | Appears automatically atop results | User opens the AI Mode tab |
| Format | Ranked list of links | AI summary + a few citations, links below | Conversational chat, ongoing |
| Interaction | One query, scan, click | Read the summary, maybe follow up | Ask, refine, follow up freely |
| Sources | Ten blue links | Multiple sources, few surfaced | Query fan-out across many sources |
| Data | Indexed pages | Gemini’s indexed crawl | Live signals, Shopping Graph, reasoning |
| Your visibility | Rank position | Cited or invisible | Cited across sub-queries or invisible |
The strategic point is the same for both: you are no longer competing only for a rank position. You are competing to be the source the AI quotes. If it does not name you, you are invisible even when your page is technically ranking well.
Do AI Overviews reduce website traffic?
I will be honest rather than reassuring, because you deserve the real number.
Yes, AI Overviews reduce clicks to websites on the queries where they appear, and the effect is significant. The most rigorous data comes from Pew Research, which tracked nearly 69,000 real searches and found that users clicked a traditional result on only 8% of searches when an AI Overview was present, compared with 15% when it was not. Roughly halved. Pew also found that people click a link cited inside the Overview itself only about 1% of the time.
A separate field study found AI Overviews cut outbound organic clicks by 38% on the queries that triggered them, with zero-click search rising from 54% to 72%. AI Mode is more extreme again: some analyses suggest the large majority of AI Mode searches end with no click to any external site at all.
This is the zero-click search era arriving in force. For years “zero-click” meant weather boxes and currency conversions. Now it means genuine, considered answers to the kind of research questions that used to send people to your blog.
Two things stop this from being a catastrophe. First, the pain is uneven. Informational and top-of-funnel content takes the heaviest hit; commercial, branded, and transactional queries where someone wants to buy, book, or contact are far less affected, because a summary cannot complete the purchase for them. Second, there is genuine upside to being cited: some data suggests pages cited inside an AI Overview can earn more clicks than a standard first-place result would have, because the citation acts as an endorsement. The traffic is not gone. It is consolidating around the sources the AI trusts.
How do I show up in Google AI Overviews and AI Mode?
This is where I spend most of my time with clients, and the good news is that the work is concrete.
There is no special markup, secret file, or “AI Overviews” toggle. You earn citations by making your content genuinely easy to extract and genuinely worth trusting. That combination, extractability plus authority, is the whole game. It has a name now: answer engine optimisation (AEO), sometimes called generative engine optimisation (GEO). I have written the complete AEO and GEO playbook as a standalone guide, and if you read one thing after this, read that. Here is the condensed version.
1. Answer first, elaborate second. Put a one or two sentence direct answer immediately under each heading, then explain underneath. AI systems lift that top block. If your answer is buried in paragraph six after a personal anecdote, it will not be extracted. Write the way this article is written: the bolded sentence under each question is not a stylistic tic, it is bait for the extractor.
2. Structure for machines and humans at once. Clear H2 and H3 headings phrased as real questions. Short paragraphs. Bulleted lists. Comparison tables, because AI systems love structured comparisons and lift them directly. The table earlier in this article exists for exactly that reason.
3. Add schema, especially FAQPage and Article. Structured data gives Google’s AI machine-readable signals about what your content means, and pages with comprehensive schema get cited noticeably more often. FAQPage schema on question-and-answer content is close to non-negotiable, because it hands the AI pre-formatted question-answer pairs. If you are unsure how, this is a one-time technical job worth doing properly across your whole site.
4. Publish something the AI cannot get elsewhere. Google’s AI selects sources on expertise, extractability, and uniqueness. Content with original statistics sees materially higher visibility in AI answers because it gives the model a verifiable claim to cite. Your own data, your own client results, your own tested opinion. Rehashing the same ten sources everyone else rehashed gives the AI no reason to pick you over them.
5. Build authority beyond your own domain. Citation likelihood tracks with your presence across the wider web. Notably, LinkedIn climbed to the second most-cited source for AI answers in early 2026. Where credible people discuss your topic, and whether your name appears there, feeds the model’s trust signals.
None of this is exotic. It is disciplined fundamentals, applied with the extractor in mind rather than only the human reader.
What is the Australian context?
If you are running a business here, the timeline is not theoretical.
AI Overviews have been live in Australia since October 2024, and AI Mode has been rolling out through tier-one English markets including Australia across 2026. Google confirmed the Australian AI Overviews launch itself, and Australian users have been unusually quick to adopt AI search, with local coverage noting Australia among the leaders in AI query volume per person.
Practically, that means you cannot treat this as a US phenomenon to monitor from a distance. Your Melbourne, Sydney, and Brisbane customers are already getting AI-written answers to questions your website used to answer. Waiting for it to arrive is not a strategy, because it arrived.
What businesses should actually do this quarter
Let me turn all of this into a short list you could hand to your team on Monday.
- Segment your keywords by intent. Accept traffic loss on pure informational terms, and defend hard on commercial, comparison, and branded queries where a click still leads to money.
- Rewrite your top ten pages answer-first. Direct answer under every heading, tables where a comparison exists, tight paragraphs. This alone shifts citation odds.
- Deploy FAQPage and Article schema sitewide. A one-off technical project with lasting returns.
- Commit to original data or genuine expertise in every new piece. No unique contribution, no citation.
- Track impressions and citations, not just clicks. In Google Search Console, rising impressions with flat clicks is not failure, it is the new baseline. Measure whether you are being surfaced.
- Strengthen off-site presence, LinkedIn in particular, so the model sees your expertise corroborated elsewhere.
The businesses that win the next two years are not the ones fighting to keep the old click patterns. They are the ones becoming the source the answer is built from. Search did not disappear. It changed shape, and the shape rewards genuine authority delivered in a format machines can lift. That is a game a serious business can still win.
If you want to go deeper on the mechanics of getting cited, my full AEO and GEO guide is the next read, and if you are auditing your wider toolkit, the best AI tools for Australian businesses in 2026 covers what I actually use.
I am Amjid Ali, an AI and technology leader based in Melbourne. If your organic traffic is sliding and you want a straight answer on what to do about AI search, get in touch.