Two months later, on 20 May 2025, Google rolled out AI Mode to everyone in the U.S. Just a couple of months later, Google went full steam ahead and rolled it out to India, later the UK, and now 180 more countries and territories.
Like it or not, search has changed.
Here’s everything we know about Google’s AI Mode to date:
Explore our index of 4.6 million (and rapidly growing!) AI Mode queries and responses in Ahrefs Brand Radar:

Google’s AI Mode turns your traditional Google search into a conversational experience.
In classic Google search, Google shows a list of links for your query. You then have to click through one or multiple links to get your answer.

If you’re unsatisfied with what you’ve found, you’ll have to make multiple searches.
In AI Mode, Google uses their Gemini 2.5 generative AI model to generate an answer for your questions and a list of links where they’ve cited from.

You can then follow up by asking more questions.

AI Mode is multimodal, meaning that you can type, talk, snap a photo, or upload an image to get your answers.
If you’re familiar with AI assistants like ChatGPT or Perplexity, AI Mode works in the exact same way.
Google’s eventual goal is to make AI Mode agentic, i.e. it can do things for you.
For example, on 21 August 2025, Google added agentic booking features to AI Mode for Google AI Ultra subscribers. Users will be able to make a request, e.g. “find me a dinner reservation for 4 people this Friday at 7 pm” and AI Mode will be able to interpret the request, check real-time availability, and link you to the booking page.

AI Mode is available as a tab on Google search.


You’ll be able to access it if you’re in one of the 180+ countries and territories that Google has rolled AI Mode out to. (See the list of supported countries and territories here.)
Right now, AI Mode is only available in English.
In classic search, Google takes your query and shows you a list of links that is as relevant as possible to what you’re looking for.
In AI Mode, Google uses a query fan-out technique where it takes your query/prompt and divides it into multiple sub-queries.
For example, a prompt like “Could you suggest Bluetooth headphones with a comfortable over-ear design and long-lasting battery?” could generate subqueries like:
- Best Bluetooth headphones with over-ear design
- List of top-rated over-ear Bluetooth headphones with long battery
- Most comfortable over-ear Bluetooth headphones
- Bluetooth headphones with longest battery life
- Which Bluetooth over-ear headphones charge fastest?

Source: Aleyda Solis
The system then retrieves Google’s search results based on these subqueries. Unlike classic search where each webpage is an answer, the system lifts relevant chunks of information (e.g., paragraphs, tables, images) from each source and synthesizes them into one answer.
It then uses grounding to match the generated answer to citations.

Source: iPullRank
Before AI Mode, Google had already rolled out AI Overviews: AI-generated summaries for search queries.
Based on our research, AI Overviews reduce clicks by 34.5%.

We can expect this to be even worse with AI Mode. After all, users can now get answers by continuously asking questions without having to click on any links.
In fact, early research done by SEO agency iPullRank showed that only 4.5% of AI Mode Sessions result in a click.
Sundar Pichai, the CEO of Google, confirmed in an interview with Lex Fridman that AI Mode is the future of search. So, we all have to live with this zero-click reality.
If you’re looking for a shred of good news, it’s that AI Mode’s adoption has not taken off so far:
- A study by iPullRank showed that over 50% of users tried AI Mode once and didn’t return. And only about 9% of users used it 5+ times, which indicates that early adoption is low.
- Another study by Aleyda Solis, this time focused on adoption in the UK, showed that adoption spiked when AI Mode was first introduced and has since started to decrease.

Unfortunately, Google has not made the data for AI Mode available. Right now, it gets combined with regular search in Google Search Console.
And it does not seem like they intend to separate it and show it to us. For good reason: It’s going to obviously show that every site is losing clicks, which goes against their current position of “AI search isn’t killing search clicks.”

So, to track your brand’s presence in AI Mode, you’ll have to use a tool like Ahrefs’ Brand Radar.
Enter your brand in Brand Radar, go to AI responses, select AI Mode, and you’ll see the queries your brand is appearing for.

You can also enter your competitors and see how you’re comparing with their presence in AI Mode.

Not only does LLMs like Google’s Gemini not reveal how they select sources, the links/citations in AI Mode are different every time you search.
So the truth is: no one truly knows how to be a cited website in AI Mode. It’s all theory at this point.
But that is not to say there aren’t any ideas. There are some strategies, based on studies done by ourselves and other companies on LLM optimization. We can guesstimate that these strategies will lend themselves to AI Mode too.
1. Check that AI Mode is saying accurate things about your brand
LLMs can hallucinate. Google’s Gemini is no exception. So, you’d want to make sure that whatever AI Mode is generating about your brand is correct and accurate.
To do this, enter your brand into Brand Radar to see what responses AI Mode is generating about your brand.

Check that they’re generating the correct information about your brand. If they’re not, note down the queries where errors are appearing, what’s wrong, and what sources were cited.
Sometimes, these errors appear because you have inconsistent brand assets like logos and taglines. You’d want to fix this by checking your website and social profiles for consistency.
I highly recommend following Despina’s brand SEO process to do this.
Another way AI generates incorrect information is because there’s too little information on your website. You may need to close such content gaps. That’s one of the reasons why we published the post “How to pronounce Ahrefs”.

Follow this workflow to see if you need to publish a new page or section to fill gaps:

If the error is due to the fact that AI is incorrectly citing other sources, you’d want to consider updating your content or brand assets on third-party sites like YouTube, G2, Google Business, and more.
2. Boost your brand’s presence online
In our study of 75,000 brands across millions of AI Overviews, we found that branded web mentions had the strongest correlation with brand mentions in AI Overviews.

Makes sense: Most LLMs scrape the Internet to train their models, so the more your brand is mentioned online, the more likely LLMs can associate you with relevant topics.
You’d want to build brand awareness by:
- Getting featured in “best of” lists (e.g., “best AI SEO tools”) and industry roundups
- Encouraging customers to mention your brand in reviews and case studies
- Actually building a great product that people share via word-of-mouth without prodding or prompting
- Making unforgettable content that people will share, whether that’s impactful storytelling, important data, lived experiences, or real-world hard-earned insights
- Running digital PR campaigns and getting mentioned on authoritative sites
For example, you can find collaboration and PR opportunities by going to Ahrefs Brand Radar and clicking on the Cited Domains report.

In this example, we could potentially pitch to two authoritative publications: Tech Radar and Forbes.
3. Promote your content on UGC sites
When we looked at the top 50 sites cited in Ahrefs Brand Radar for Google AI Overviews, we saw that AI Overviews leaned heavily on UGC sites like Reddit, Quora, and YouTube.
Makes sense: YouTube is Google’s own property and Google also signed partnership deals with Reddit. I can see why they would favor those sites for AI Overviews and I see no reason why it would be different for AI Mode.
For example, even though we didn’t start our YouTube channel to rank in AI search, we’ve benefited from publishing consistently on YouTube since 2018. Our keyword research tutorial is one of the cited sources for “how to do keyword research” using AI Mode:

In our inaugural Ahrefs Evolve conference, Sam Oh, our YouTube maestro, shared his best YouTube tips:
- Add target keyword to your title and description
- Mention relevant keywords in your content
- Cover necessary subtopics in your content
- Use timecodes or chapters with relevant keywords
- Hold people’s attention as long as possible

I also recommend watching his YouTube video that walks you through our entire YouTube strategy:
Reddit is a hard platform to crack as Redditors are well-known to hate blatant marketing. The only way to do well is to be an active Redditor yourself. Leave useful comments, participate in discussions, share valuable content, and become a known person in your niche.
Even though it’s a few years old now, I wrote a guide on how to actually market on Reddit the authentic way. I recommend reading it.
4. Use the Atomic Content workflow to structure LLM-friendly content
SEOs have known for years that structuring content in a certain way can help a page rank in featured snippets. LLMs are the same too—they like content to be structured in a certain way.
At Ahrefs, we call this atomic content: self‑contained sections within larger documents that act as indivisible units of knowledge.

Each atomic unit should be able to stand on its own, delivering a complete answer even if extracted and surfaced by Google, ChatGPT, Perplexity, or other AI search platforms.
To create atomic units, you’d want to apply the Bottom Line Up Front (BLUF) framework. Simply put: Start each article and each section within your article with a direct answer or clear statement that fully addresses the core topic.
Here’s an example from one of our articles:

Using BLUF is ideal because it works for both humans and LLMs.
Despite your greatest wishes, humans scan first before reading. BLUF gives them the answer right away and allows them to decide if they want to continue. BLUF also makes your content instantly understandable and retrievable, which to a LLM, is a dream come true.
5. Keep your content fresh
Our study of 17 million citations across seven AI search platforms showed that AI assistants prefer citing fresher content.
Even though Google is the least influenced by content freshness, Gemini—which powers AI Mode—seems to work differently and prefer fresher content:

So, our research suggests that keeping your content updated could be useful in being cited by AI Mode.
You can find content worth updating by entering your domain into Ahrefs’ Site Explorer, going to the Top pages report, and setting a filter for declining traffic over a date range (typically six months to a year.)

Scan the Content changes column for “minor” or “moderate” changes as they indicate that these are content yet to be refreshed by your team.

Then, follow our guide below to update your content.
6. Build long-tail keyword clusters
Since AI Mode “fans out” your prompt into multiple long-tail sub-queries, optimizing for long-tail keywords can increase your odds of being cited in AI Mode.
This is a strategy most SEOs were doing prior to AI Mode, which goes to show that despite the naysayers, SEO is not dead and is still one of the best ways to rank high on both SERPs and AI search. GEO is mostly SEO after all.
To get long-tail keyword ideas, you can perform a competitor gap analysis in Ahrefs Brand Radar.
Enter your brand and competitors, hover over AI Mode, then click on the number in Others only for your brand.

This will show you the prompts and queries that your competitors are visible for that you’re not. Study the prompts for long-tail content ideas.

Final thoughts
Will AI Mode really become the default Google search experience in the future? Google’s CEO Sundar Pichai has hinted that it’s a yes, and we’ve also seen Google going full throttle by rolling out AI Mode rapidly.
But some SEOs think otherwise. They think Google is merely testing to see if people will use AI Mode. This is why Google has put AI Mode as a separate tab, rather than ‘force’ users to use AI Mode only. And so far, as we’ve seen, the adoption rate has been low.
Personally for me, I think AI Mode will be the default search experience, with web search as a separate tab.
The genie’s out of the bottle and you can’t put it back ever again. Many people, myself included, have defaulted to using conversational chatbots as our main search engines.
It’s just a better user experience than having to sieve through hundreds of links and still not find what you want.