Gemini 3 Now Powers AI Overviews, and Follow-Ups Open AI Mode

Maybe you’ve noticed AI Overviews reading differently this week, or a new prompt under them to keep asking questions. You’re not imagining it. The Gemini 3 AI Overviews update changes the model behind those answers. It also adds a new path from an AI Overview straight into AI Mode. For site owners, the short version is this: nothing about eligibility has changed, but how people move through your topic on the results page has.

Below I cover exactly what Google announced, what it means for searchers, what it means for your site and your Search Console data, and what I would watch over the next few weeks.

What Google Announced on 27 January 2026

On 27 January 2026, Robby Stein, VP of Product for Google Search, published a post on Google’s Keyword blog. It announced two changes to Search. You can read it in full in Google’s announcement of the AI Overviews and AI Mode updates.

• Gemini 3 is now the default model for AI Overviews, worldwide. The same model family that has powered AI Mode since November 2025 now writes the AI answers at the top of the results page, wherever AI Overviews are offered.
• Follow-up questions continue into AI Mode. After reading an AI Overview, a searcher can ask a follow-up and move into a back-and-forth conversation in AI Mode. The context of the first AI Overview carries over, so they don’t have to start again.

Google says the handoff from AI Overviews into AI Mode is available on mobile, worldwide. The post doesn’t say when, or if, the same flow will come to desktop. So for now I treat it as a mobile behaviour.

Google also said its testing showed people prefer an experience that flows naturally into conversation. That matches what it tested on mobile in December 2025. Back then, an option to keep asking appeared at the bottom of expanded AI Overviews.

How This Fits Google’s Earlier Gemini 3 Rollout

This is the next step in a series that started in November. On 18 November 2025, Google brought Gemini 3 to AI Mode on the day the model launched. It started with paying Google AI Pro and Ultra subscribers in the US.

At that time Google said it would start sending harder questions in AI Mode and AI Overviews to Gemini 3, while simpler questions stayed on faster models. The January change goes further. Gemini 3 is now the default for AI Overviews for everyone, not just for some complex queries or subscribers.

What Changes for Searchers

For the person typing a query, three things are different.

1. AI Overviews Should Handle Harder Questions Better

Google calls Gemini 3 its strongest reasoning model. In practice, that should mean AI Overviews cope better with multi-part questions, comparisons and queries with limits, such as “best way to heat a small garden office without mains gas”.

Google also uses a technique it calls query fan-out. The system breaks a question into related sub-searches across subtopics and data sources. Then it pulls the findings together. Google said in November that Gemini 3 improves fan-out, so the system can find useful content it missed before.

2. One Question Can Now Turn Into a Conversation

Before, an AI Overview was a dead end. You read it, clicked a link, changed the query, or opened AI Mode separately and started again.

Now a searcher on mobile can ask “what about for a rented flat?” right under the AI Overview. They carry on in AI Mode with the first context intact. For a lot of everyday research, I expect that to keep people inside Google’s AI experience for longer.

3. AI Overviews Still Do Not Show for Every Query

The model change does not mean AI Overviews appear everywhere. Google’s own guidance says they only show when its systems judge they add something to classic results. Plenty of navigational and simple queries will still show a normal results page.

What Changes for Site Owners

This is where most of the questions come from. So let me be direct about what is and isn’t different.

Eligibility Has Not Changed

There is no new requirement to appear in AI Overviews or AI Mode. Google’s documentation on AI features and your website is clear. A page needs to be indexed and able to show in Search with a snippet. There are no extra technical requirements and no special tweaks for AI features.

That means:

• No special schema markup gets you AI Overview citations.
• No “AI text file” or new machine-readable file is needed.
• The same controls still apply. nosnippet, data-nosnippet, max-snippet and noindex limit what Google can show from your pages, including in AI features.

Here’s a point I often see people mix up. The Google-Extended token controls use of your content for training and grounding in Google’s other AI systems, such as Gemini apps. It is not the switch for AI Overviews in Search. If you block Googlebot or use noindex, you are opting out of Search as a whole, not just the AI parts.

You may be tempted to restructure pages just for AI. But Google has also said it doesn’t want content chopped into tiny pieces for machines. I covered that in what Google said about chunking content for AI, and it applies just as much under Gemini 3.

Which Pages Get Cited May Shift

Google has not said that citation patterns will change. But when a different model writes the answer and fans out to different sub-queries, the set of supporting links can shift. A page that was often linked from an AI Overview last month may appear less. A page that never featured may start to.

This is not a penalty or a ranking update. It’s a change in how answers are put together. My advice: don’t react to a few days of data by rewriting pages that were doing well in classic results.

Link Display: What to Expect

Google describes the new flow as one experience with clear links to keep exploring. In December 2025 Google also said it was adding more inline links in AI Mode. It was also adding short notes that explain why a link might be worth visiting.

Put together, this is the realistic picture for a site owner:

• Your page might be linked in the first AI Overview, in a later AI Mode follow-up, or in both.
• Links in AI Mode can sit inline within the answer, not only in a side panel or a list at the end.
• Gemini 3 in AI Mode can build richer layouts, such as tables and grids. So where a link sits on screen may change from one query to the next.

What hasn’t changed is that links come from pages Google has indexed and finds relevant. In my experience, clear, well-organised pages that answer specific questions are still the best way to be one of those links. I go through Google’s own tips in my post on succeeding in AI search.

How Gemini 3 AI Overviews Show Up in Search Console

This is the part that will matter most when you report to clients or your boss. Search Console has no separate AI Overviews or AI Mode report. Traffic from both is included in the Performance report under the Web search type, mixed in with classic results.

Google’s Search Console help explains how each one is counted. The table below sums it up.

That last row is important for this update. When someone asks a follow-up from an AI Overview and lands in AI Mode, Google treats it as a new query. Any impression or click your page earns in that follow-up answer is recorded against the follow-up query. It is not recorded against the first search.

In practice, expect to see more long, chatty and sometimes odd-looking queries in your Performance report over time. Some of those will be follow-ups typed after an AI Overview. Don’t brush them off as junk. I find they show what people want to know next.

Also keep in mind that Search Console doesn’t include data from Search Labs experiments. So anything you saw in a Labs test won’t show up in your reports.

If you want cleaner trend lines while this settles, the newer weekly and monthly views help smooth out daily noise. I explained how to use them in Search Console’s weekly and monthly views.

How to Check the Impact on Your Site

AI Overview data is blended into Web search, so you can’t isolate it with a single filter. But you can build a fair picture with a short routine. This is the one I use.

1. Set a clean comparison window. In the Performance report, compare the weeks after 27 January with the same number of weeks before. If you can, avoid windows that overlap the December 2025 core update, so you aren’t mixing two changes.
2. Split by device. The AI Overview to AI Mode handoff is mobile only for now. Filter by Mobile and then by Desktop. A change that shows up on mobile but not desktop is more likely linked to this update.
3. Focus on informational queries. Use a query filter or regex (a text pattern) for question words such as how, what, why, best and vs. These are the searches most likely to trigger AI Overviews.
4. Watch impressions and CTR together. Rising impressions with falling clicks on question queries can mean your pages are shown in AI answers, but the question gets answered on the results page. Falling impressions on those queries can mean citations have moved to other sites.
5. Look for new conversational queries. Sort queries by impressions. Look for longer, follow-up style wording that didn’t appear before. Note them as content ideas.
6. Check the live results yourself. Search your ten most important informational queries on a phone. Read the AI Overview, ask one or two follow-ups and note which sites are linked. Repeat every couple of weeks.

Keep a simple log of what you see. Model updates don’t come with a completion date on the Search Status Dashboard. So your own notes are the best record of when things moved.

What to Watch Over the Coming Weeks

Gemini 3 AI Overviews are new, and Google has left a few things open. These are the points I would keep an eye on.

• Desktop. Google only confirmed the follow-up handoff on mobile. If it arrives on desktop, expect a second round of changes in desktop behaviour.
• Citation spread. Notice if AI Overviews in your niche link to more sites, fewer sites, or different types of sites, such as forums, manufacturers or publishers.
• Click quality. Google says clicks from results pages with AI Overviews tend to be higher quality, with people more likely to spend time on the site. Check that against your own engagement data in GA4. Don’t just take it on trust.
• Personalisation. Google has been adding personal context to AI Mode. As more conversations start from an AI Overview, those personal signals could shape more follow-up answers. I wrote about that in Personal Intelligence in AI Mode.
• Your own brand queries. Search your brand name plus common questions, such as “is [brand] any good” or “[brand] vs [competitor]”. Make sure the information Google can find about you is correct and current on your own site and Business Profile.

What to Do Now, and What Not to Do

In my view, the sensible response to this update is mostly about keeping the basics sharp. You don’t need to reinvent your content.

Do

• Make sure important pages are crawlable, indexed and allowed to show a snippet.
• Answer the obvious next question on the page. If your guide explains how to do something, cover cost, time and common mistakes too. Those are the likely follow-ups.
• Keep key content in text, with images or video where they really help.
• Keep structured data accurate and matching what is visible on the page.
• Use internal links so Google can find related pages that answer follow-up questions.

Do Not

• Don’t add special markup or files that promise AI visibility. Google says none are needed.
• Don’t apply nosnippet sitewide to “protect” content. It also removes your snippets from classic results.
• Don’t rewrite top-performing pages based on one week of data.

Conclusion

Gemini 3 AI Overviews don’t change the rules for getting into Google’s AI answers. Your pages still need to be indexed and snippet-eligible. All of the traffic still lands in Search Console’s Web search type. What has changed is the journey. A stronger model writes the first answer. On mobile, that answer now leads straight into an AI Mode conversation, where every follow-up counts as a new query.

My suggestion for this week: set a before-and-after comparison in Search Console, split by device. Then start a simple log of which sites appear in AI answers for your key questions. If you haven’t already, it’s also worth helping loyal readers find you more easily through Google Preferred Sources.