Google I/O is Google’s annual developer conference, where Google unveils its latest products, shares major updates about new products and technology, and connects with the global developer community through keynotes, demonstrations, and hands-on sessions.
Each year, Google I/O is an important event for search marketers to learn about Google’s plans for updating and improving its search and AI products, and it usually includes major announcements and changes to Search that go into effect that same week.
Jump To:
Google I/O 2025 announcements: Google pushes forward with Generative AI
It wasn’t surprising that Google’s new products, ‘AI Mode‘ and ‘AI Overviews,’ were the centerpiece of Google’s search announcements at this week’s Google I/O. These products are now both powered by Google’s latest Gemini model, Gemini 2.5, which Google claims will lead to more accurate and engaging generative AI answers.
AI Overviews, which launched publicly in 2024, present AI-generated summaries at the top of search results for specific queries. These AI answers synthesize information from multiple sources, offering direct answers within Google’s search results. They also include links to the cited websites and, in some cases, links back to Google’s own search results.
At Google I/O, Google reported that AI Overviews have led to a 10% increase in usage for queries where they appear, especially for more complex, multi-part, or multimodal questions. Google also asserts that its users are engaging more deeply, utilizing AI Overview answers and asking follow-up questions. As a result of this positive user feedback, Google announced a major expansion of AI Overviews in Search, making them available in over 200 countries and territories and supporting more than 40 languages.
One of the most significant announcements related to SEO at Google I/O was that AI Mode is now public and available to all Google searchers in the U.S. AI Mode is now a separate tab at the top of Google Search — currently positioned to the left of all other Search tabs (including “All” search results).
AI Mode presents a “re-imagining of Search” by allowing users to ask highly detailed, multi-part, or follow-up questions in a conversational, chat-like interface. AI Mode uses Google’s proprietary “query fan-out” technique to break down complex queries into subtopics, issuing multiple searches in parallel and synthesizing the results into comprehensive answers.
AI Overviews and AI Mode have been at the forefront of many of Google’s public communications over the past year or two, so many SEO professionals were unsurprised to see Google make these official announcements. Google had previously launched AI Mode as an experiment in beta earlier this year, and our SEO team has been testing it extensively ever since, especially as it relates to linking out to external websites:
Google also announced some new features that will be coming to Google Gemini and AI Mode, such as Deep Search (for thorough, research-intensive queries), AI-powered shopping with interactive product panels, and upcoming agentic capabilities that will make Gemini and AI Mode more personalized. By drawing on users’ data, such as search history, app usage, Google Drive, and Gmail (with their consent), Google will provide highly personalized answers in Gemini and AI Mode starting this summer. Google also mentioned that users will be able to opt out of personalization in responses.
How will AI Overviews and AI Mode impact SEO?
There is no question that these new generative AI products and new AI features in Search will have a major impact on how users engage with Google’s organic search results going forward. This has led to many questions from publishers, site owners, and SEO professionals alike. Below are some of the most significant questions that these tools might present to organic traffic and visibility, and the most common questions asked of Google as it rolls out these new products:
How can marketers track the impact of AI Overviews and AI Mode?
One of the main challenges that the SEO industry shares about AI Overviews and AI Mode is the inability to directly track these products in Google Search Console or Google Analytics. Unlike other Google Search features, like Google Images, which have their own separate filters in Google Search Console, there is currently no such filter for AI Overviews, even though they have been displayed in search results for over a year. AI Overviews also do not have any type of dedicated performance report in Search Console, like the breakout report for Google Discover.
Similarly, AI Overviews cannot be broken out in Google Analytics 4. While some AI Overviews carry over ‘Scroll to Text’ highlighting into the URL, which can be tracked in Google Analytics 4, other Search features like People Also Ask and Featured Snippets also carry over this highlighting, which means Scroll to Text highlighting from all of these Search features would be combined in GA4 reporting.
Similarly, a brand new study by Ahrefs shows that AI Mode traffic is not showing up in Google Search Console and cannot be broken out in Google Analytics 4. AI Mode does not send its own referrer, so the traffic is currently grouped under “Direct” or “Other” in GA4. However, John Mueller from Google has since stated that this is “a bug” that Google is working to fix – so we should have AI Mode tracked as referrals from Google in Analytics soon (and hopefully labeled as organic traffic!)
Also, Google’s documentation does state that AI Mode traffic will appear in Google Search Console, grouped in with the other clicks and impressions shown there (without dedicated reporting). John Mueller also confirmed that AI Mode clicks and impressions will begin to populate in Google Search Console soon, but it is unlikely that we will see this traffic broken out in any type of dedicated filter or report for AI Mode.
The SEO industry has consistently asked Google to provide dedicated tracking for AI Overviews (and now AI Mode) in Google Search Console. At a recent Google Search Central meetup in New York, this was by far the most asked question during the Q&A section. SEO professionals and site owners can’t fully understand the impact that AI Overviews are having on their organic traffic and visibility without dedicated reporting.
How will AI Overviews and AI Mode affect organic traffic?
Over the past year, Google has consistently stated in public discussions about AI Overviews that:
“With AI Overviews, people are visiting a greater diversity of websites for help with more complex questions. And we see that the links included in AI Overviews get more clicks than if the page had appeared as a traditional web listing for that query. As we expand this experience, we’ll continue to focus on sending valuable traffic to publishers and creators.”
However, new industry studies have tended to produce outcomes that tell a different story from Google’s official statements about traffic from AI Overviews. Several new SEO studies that use first-party data, including our own study by Amsive’s SEO Strategist, Will Guevara, have shared that the presence of AI Overviews does indeed cut into organic traffic — which isn’t entirely surprising, given that AI Overviews are designed to answer the user’s question directly in Google’s search results. Other studies have shown traffic declines resulting from AI Overviews to be even more dramatic. The SEO Director at Mail Online, Carly Steven, shared with Press Gazette that AI Overviews have resulted in clickthrough rates that are “56.1% lower on desktop and 48.2% lower on mobile.”
Another organic search marketing challenge is that AI Mode appears to be structured to encourage users to stay within Google’s ecosystem as much as possible. The responses from AI Mode often include several thumbnail links to external websites, along with an occasional link icon, as highlighted below:
These link icons point to a list of organic search results, from which the user can potentially jump off to visit external websites. Occasionally, AI Mode includes external links within the answer using a solid black underline, which sends users to an external website in a new tab. In many other cases, AI Mode uses dotted line links to link to Google’s own products, such as a Google Shopping or Google Business Profile result, or even to a brand new search result. These design updates work effectively to keep the user within AI Mode, getting all the information they need directly within Google’s own interface — which will likely lead to organic traffic declines for many types of websites.
For sites that monetize through traffic, such as by using display ads, affiliate links, and/or subscriptions, this drop in traffic will have meaningful business impacts. Indeed, these concerns about organic traffic are now reaching mainstream publications, like NY Mag’s ‘Intelligencer,‘ which reported on the traffic impacts of Google’s AI products. Many publisher sites will need to explore new ways to drive traffic, such as with Google Discover, email marketing, social media, multi-modal approaches to driving organic visibility, and brand awareness strategies to drive direct traffic.
Are responses from AI Overviews and AI Mode accurate?
Another question about Google’s new AI products is the accuracy of their responses.
At Google I/O, Google announced that they are bringing “a custom version of Gemini 2.5, our most intelligent model, into Search for both AI Mode and AI Overviews in the U.S.” This announcement was likely made in response to the many popular examples of AI Overviews getting answers blatantly wrong. Although “hallucinations” are a characteristic of large language models, Google is confident that its new AI products will increasingly provide more accurate answers, as noted in this Wired article:
“The accuracy rate for AI Overviews is on par with other well-established Search features like featured snippets, which have been providing helpful information for a decade,” says Craig Ewer, a spokesperson for Google. “We continue to make further gains in areas like factuality, and these improvements inform how we approach AI Mode.”
However, anecdotal examples of AI Overviews and AI Mode getting information incorrect appear to surface across social media almost every day. One way that these AI products currently get information wrong is that they tend to display inconsistent information with what appears in Google Maps:
When asked about inconsistencies between AI Overviews and Google Business Profile (Google Maps) at the Google Search Central meetup in New York, Google did not yet have an answer for when these products might be integrated, but took note of the feedback.
It’s fair to assume that as Google improves these products over time, we will see more consistency between Google Business Profile attributes and AI responses. But for now, it might be a good idea to use LLM tracking tools, such as ZipTie, Ahrefs, Profound, or Peec AI, to monitor answers in AI Overviews for any potential inconsistencies with your Google Maps data.
Thinking About the Future of SEO
SEO has gone through many changes over the years. However, Google’s AI Overviews and AI Mode present a unique challenge, as these tools are highly effective at keeping organic searchers within Google’s ecosystem.
Nevertheless, these developments present new opportunities that will shape the future of SEO. They also reflect how SEO continues to evolve—from a siloed technical discipline to smart, holistic marketing. Today, earning visibility means building presence across the entire web, not just in traditional search results. The tactics below highlight how we’re approaching that shift.
1. Drive mentions/references in LLM (AI) responses
The biggest SEO opportunity right now is adapting our approach to drive visibility across the new platforms where searchers are looking for answers. Although ChatGPT’s market share may still be small in comparison to Google, it’s true that consumers are rapidly adopting Large Language Models as the places to start their search. For this reason, we have shifted our SEO approach to focusing on driving visibility, mentions, and positive sentiment across LLMs like ChatGPT, Google AI Overviews & Gemini, Claude, Perlplexity, Copilot and more.
While ranking well in traditional results still matters — for example, sites that rank #1 in Google’s blue links are used as a source in AI Overviews about 25% of the time — it’s no longer a guarantee of inclusion. That’s why we focus on optimizing for visibility across a broader set of signals and platforms, not just rankings.
Our partnership with Profound and integration of tools like ZipTie and Ahrefs give us the data and insights we need to understand how to help sites appear in AI responses. While these LLMs may drive less traffic to websites, being mentioned in their answers is essential for businesses to drive awareness and visibility. We also find that traffic from LLMs tends to come from more engaged users who are more likely to convert, as they are likely to be further along in their decision journey after their conversations with LLMs.
2. Understanding the value of LLM visibility & traffic
Search is changing, and organic traffic for many queries might be on the decline, but that doesn’t mean search is no longer valuable. In fact, it is now more important than ever to ensure that your brand, your products and services, and the experts and leaders who work at your company are visible and presented in a positive light.
Large language models help users research the right resources for them, and as this process becomes more personalized, it’s likely that this will help users find the right brands, products and services for their specific needs. A big part of this is ensuring that you have the content that helps answer their questions and showcases your brand as an expert across all the different touchpoints the user might use during their search journey.
Conversations on LLMs could replace the traffic that previously flowed to upper-funnel, informational content, such as blog articles – but they could lead to higher-quality visitors who are more interested in inquiring about your products and services. This is why showcasing the E-E-A-T (experience, expertise, authoritativeness and trustworthiness) of your brand and the people at your brand is essential – LLMs can reference your expertise throughout the discovery process.
3. Use multi-modal AEO to diversify asset visibility
Search has advanced well beyond just textual content. Users are increasingly seeking answers through visual searches, such as Google Lens, which currently handles 20 billion searches per month and is gaining popularity among younger users. While image optimization has been at the core of our SEO offerings for years, we are now even more excited to help clients reach new customers with visual assets that are well-optimized for AI search using answer engine optimization (AEO).
4. Amplify brand discovery via Google Discover
Google recently announced that its content discovery engine, Google Discover, is coming to desktop. Since 2018, Google Discover was only available on mobile devices, and our team has spent every year since crafting and honing our Google Discover consulting and auditing services. Because our SEO team works with so many publisher websites, we have extensive access to data and discoveries across dozens of high-performing Google Discover accounts, giving us a unique view into which tactics work to drive Discover traffic over time. We’ve seen that publishing a well-optimized article for Google Discover can drive millions of clicks to a single article, often driving more visibility than the article would have received from organic search alone.
Google adding Discover to its desktop search engine is welcome news for publishers looking to drive more organic traffic to their content.
5. Develop cross-platform SEO visibility
SEO has evolved beyond simply focusing on website rankings. The truth is that search has become cross-platform, as users start their searches and explore new information across social platforms like TikTok, LinkedIn, YouTube, X and Bluesky. We have continuously tested this by leveraging different social media platforms like LinkedIn to see whether that content would be picked up in AI responses, with promising results:
We also work with Profound to understand which LLMs are citing external websites most frequently. The below slide is from my recent keynote presentation at BrightonSEO, which shows the external websites different LLMs cite the most.
For example, Open AI (ChatGPT) most frequently cites Wikipedia; Perplexity frequently cites Reddit and YouTube; Google’s AI Overviews cite YouTube, and Microsoft Copilot cites Forbes and Gartner.
These learnings can help us ensure our clients appear across all the platforms and citations most heavily mentioned and leveraged by large language models and that their products and services are a key part of the discussion in relevant communities. We also use tactics like digital PR, plus YouTube and podcast optimization, to ensure information and answers are well-optimized for AI discovery.
The rise of AEO: Answer Engine Optimization
Learn how Amsive is tracking visibility for brands with answer engine optimization.
We’ll share high-impact strategies for visibility in answer engines (like AI Overviews, ChatGPT, Perplexity, and more) in our upcoming webinar, Accelerate Growth with Answer Engine Optimization, on June 24. Join the discussion with our SEO and AEO leaders.
