By Lily Ray
I spent last Thursday, March 20, 2025, at the Google Search Central meetup in NYC, where members of the Google Search and Search Relations teams offered a window into how they’re thinking about the future of Search, especially in light of recent AI developments.
I also live-tweeted the entire event on both X and Bluesky, so if you want the play-by-play, you can start there. But I also wanted to take a step back and summarize some of the most important takeaways, along with my honest opinions and reactions about the event.
Google is still listening. But it doesn’t have all the answers.
To start, I want to give credit where it’s due: it was generous of Google to organize this event (not to mention host it in NYC, but I’m biased there), open the floor to questions, and make space for some tough conversations among site owners, SEOs, and other attendees. It felt that Google was trying to be earnest and transparent, and it was clear that (at least some) Googlers genuinely care about the impact their decisions have on the ecosystem.
But despite the openness, it was also clear how much confusion and frustration still exists – especially when it comes to AI Overviews.
Personally, I feel like there is a bit of a disconnect between what is happening on the Search side and the AI Overview side – Google’s Search goals often feel at odds with the impact of AI Overviews on site owners and content creators. Also, it feels like Google can’t really tell us much about how AI Overviews are truly impacting traffic patterns from organic search… it seems like it’s the same talking points over and over (‘users really like it!’) without evidence.
Some pictures from the event:
The industry is desperate for AI Overview data. Google isn’t ready to give it to us (and might not ever be).
My biggest takeaway from the entire day is that the SEO industry seems united in its desire for tracking and visibility into AI Overviews (AIO). Everyone wants to understand how this new feature affects their traffic and their clients, and how to measure it.
Meanwhile, Google continued to echo its main talking points about AIO traffic, which can be found here in this press release by Liz Reid:
“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.” – Liz Reid
Google also announced that there would be some new data in GSC, but not related to AIO: the GSC API will now provide 24-hour insights for the past 8 days, creating 192 new data points in the GSC API. So for those of us who live in GSC data every day, this was definitely exciting.
Then, during the Q&A, various SEOs expressed concerns about AIO tracking. Paul Shapiro asked for UTM parameters in AIO citations, noting that ChatGPT already does this, which I thought was a great point. Glenn Gabe pushed for any kind of AIO-specific reporting in Google Search Console.
Google’s response? Long story short: AIO tracking is not happening right now.
They acknowledged our concern, and it was clear they heard us, but they didn’t offer a timeline for when AIO data might become available in GSC – or confirmation that it ever would. Google mentioned how the product is constantly evolving, making it difficult to provide meaningful data in GSC or otherwise. They also stated that UTM parameters from AIO links wouldn’t be likely because that could be seen as providing a preference for Google’s own analytics products.
For many of us, these responses are frustrating. This is especially true for those of us who have clients demanding insights about how AIO is impacting their traffic. If AIO is going to fundamentally change how users interact with search – and in many cases, cause users to bypass traditional links altogether – site owners need visibility into that behavior. We can’t report on what we can’t see.
And if I’m being totally honest, the lack of response here from Google made me (and many others) feel like the truth is that AI Overviews don’t, in fact, lead to more traffic to more websites. They likely lead to less, on average. While it’s certainly not the case for all keywords, early industry studies like a recent study by Seer confirm that AIO results tend to lead to lower organic CTRs. Frankly, this is also intuitive; why would a search feature that answers questions directly in the results lead to more overall traffic to websites?
Personally, I would prefer Google would just be honest about this information instead of telling us this feature leads to more traffic overall.
One interested moment was when Michael King asked Google if they could help us set better expectations with clients – that SEO traffic might decrease, but hopefully with higher quality visits from AIO, more conversions, etc. Google actually seemed intrigued by this question and made a note that they’d look into that. To me, that felt like an admission that AIOs do in fact lead to less overall traffic.
AIO is still producing inaccurate (and preventable) information
When asked about hallucinated or incorrect information in AI Overviews – especially when accurate details are available in Google Business Profiles – Google didn’t have a great answer, in my opinion.
An audience member asked if/when Google planned to integrate Google Business Profile data about local businesses into AI Overviews to reduce inaccurate answers for local queries in AIO.
Google’s advice? Send them examples by tagging them on social media. Submit feedback by clicking the thumbs down button in AIO.
In my opinion, that’s not good enough. AI Overviews are currently live and impacting businesses every day (hello, 14 billion daily searches on Google). It doesn’t seem fair to ask Google searchers to correct wrong answers manually when the right information already exists in a separate Google product, where business owners have submitted the correct data to Google for years already. To me, this directly contradicts Google’s broader push toward E-E-A-T and trustworthy content in Google’s results. If AIO – a Google product – is regularly surfacing incorrect information, what does that say about the systems meant to reward expertise, experience, authoritativeness, and trustworthiness?
Structured data still matters. Even in an AI-powered world.
Despite the growing role of large language models, Google reiterated that structured data is still useful. Schema is computationally efficient, easy to read, and far more precise than what LLMs can extract on their own.
Google indicated that structured data won’t necessarily influence how LLMs are trained, but it continues to play a role in how Google understands and categorizes content. This is especially true for news, shopping, and video results: it’s still a foundational tool for visibility.
In short: keep using Schema to help Google understand website content. But it was not entirely clear that Schema provides any direct benefit for LLM visibility.
Site reputation abuse, scaled content, and the generative AI gray area
Google clarified several recent spam-related policies, including the Site Reputation Abuse (SRA) guidance and rules around scaled content abuse. One of the most important takeaways is that Site Reputation Abuse is not an algorithmically-automated process. Google is using manual actions, designed to target sites that publish third-party content purely to game rankings based on domain authority.
Google also made it very clear that in order for a site to receive a Site Reputation Abuse manual action, it must check BOTH boxes: using 3rd party content AND doing so in a way that abuses the host site’s SEO reputation.
There’s also a growing recognition by Google that freelancers are not the issue for Site Reputation Abuse penalties; it’s the platforms and publishing practices that matter. Using freelancers alone is not the problem – it’s the use of freelancers in ways that also abuse the host site’s ranking signals.
Google also further clarified that the use of generative AI for content creation isn’t inherently a problem. What matters is the intent and purpose behind the content. If it’s published primarily to generate SEO traffic, especially at scale, it may be considered a violation. But using AI to support or repurpose high-quality, original content is still fair game. This isn’t entirely surprising given that Google is also simultaneously encouraging publishers to use its LLM, Gemini.
Small sites still face an uphill battle
Google spent a notable amount of time discussing how they’re thinking about smaller publishers – and what they’re doing to support them. This is part of an ongoing discussion about the impact of the September 2023 Helpful Content Update. Google shared that they’ve received over 13,000 submissions from small business owners and site operators who feel they’ve been negatively impacted by recent updates. Danny Sullivan from Google indicated that he personally reviewed all 13,000 submissions.
Google insists that its systems don’t look at brand size or status. But they do acknowledge that recognizable brands often correlate with the kinds of signals that perform well in search. In other words, brand recognition isn’t a ranking factor, but it tends to come with other advantages that boost rankings. Danny also indicated that he had trouble understanding who was behind a lot of the content he reviewed across the 13,000 sites.
The guidance here is what it’s always been: make it clear who you are, what your site is about, and why people should trust you (hello, E-E-A-T). Create for people, not for Google. Communicate your identity, credibility, and expertise.
It’s a nice message, but I still think it’s too little too late, given how destructive the Helpful Content Update has been to so many thousands of small publishers who have been desperately trying to improve their sites in the past 1.5 years, with very few seeing meaningful recoveries.
Final thoughts
The Google Search Central meetup was a valuable event, and I appreciate the team’s willingness to engage with the community. It was also great to see so many awesome colleagues and friends in the SEO industry. But it’s also clear that we’re in a transitional moment for search – one where experimentation is happening in real time, and where many of the most impactful changes (like AIO) are still a bit confusing and vague.
The lack of reporting for AI Overviews is a major issue in the SEO industry. So is the presence of incorrect information in those responses, which I’ve shared hundreds of examples of in the past 2 years.
I’m glad Google got to hear some of our perspectives in the industry, and I hope they continue to take the concerns from site owners seriously.