On Thursday, March 20, 2025, Bill Scully and I attended the Google Search Central Live Meetup in NYC. We heard from the search relations, partners, and search team members that they would give information on recent AI developments, some mythbusting, and spend time answering attendees’ questions.
AI Answers and Impact on Businesses
The topic of the day from the attendees’ perspective was AI Overviews. Google said AI is here to stay and is especially popular with younger people.
John gave an overview of how the AI system is working and how they are working on improving it by using links (called groundings) from primary sources for information-based queries. People want to see traffic data, but Google isn’t ready to provide it yet because AI features are still rolling out globally.
They also said that users’ needs are changing with innovation demands, so they need to evolve. They needed to change the types of SERP elements to match the intent of the searcher and present the best type of asset that aligns with the search query or device.
Numerous requests were made to get AI-related data, which seemed to be the search industry’s most significant burning need. Everyone needs some tracking and visibility into AI Overviews (AIO). The community needs to understand better how AI in Google’s ecosystem impacts them, whether it positively or negatively impacts traffic and engagement, and whether there is any way to measure it.
It restated the standard party line, offering a list of the benefits without any data to support these statements, and the impact on the ecosystem.m
- “better user experience,”
- “users like it,”
- “People are visiting a greater diversity of websites,”
- “They are getting answers to complex questions.”
During the live Q&A, various attendees expressed their concerns about AIO tracking. Paul Shapiro asked if Google could use UTM parameters in AIO links, similar to ChatGPT. In response, they stated that UTM parameters from AIO links wouldn’t be likely because that could be seen as providing a preference for Google’s analytics products.
Google expressed that AIO is “constantly evolving and not rolled out everywhere,” making it difficult to provide data.
Mike King asked if Google could offer more data across topics to help search professionals better explain things to their clients. Google said it would consider it.
Spam and Search Relevance
The event highlighted Google’s ongoing efforts to improve search accuracy and manage spam while integrating new technologies like AI. They indicated that 50% of all content is now SPAM and they are doing a great job of stopping it.
Google is moving from a singular definition of spam to focus on the mechanics and creation of spam. They explained why they do it to prevent malicious and other bad code and actors, as well as to try to level the playing field.
Schema Markup in the New AI World
The question was if schema (structured data) was still of any value with the growth of AI engines?
Google’s answer was yes. Google said they do incorporate it into the Google process, and Gemini uses it. They still use it for two reasons.
- They have the infrastructure to capture and process it
- It is “computationally efficient,” meaning it is easy to read and interpret, and if tagged correctly, is more precise and takes far fewer resources than extracting and interpreting on their own.
The key takeaway was that Google has built the capability to extract, process, and leverage it in its processing cycles. Google went on to state that it is still valuable for news, video, and shopping. They used the specific point of the price schema to help ensure they captured the correct price rather than extrapolating it.
They went on to say that the more websites they adopt it, the more likely they are to use it in their systems. They also suggested that if a specific schema type is listed in Google’s Structured data help guide and is reported in GSC, then it is definitely used in their processing and you should implement it.
As many have insisted, they did indicate that it may not be used in training LLMs (Large Language Models). This statement on LLM processing and the vocalization of that group of people has made some wonder if they should stop using schema. This has always perplexed me about our industry; even if it doesn’t directly benefit SEO, they want to remove it despite having no negative impact and disregarding that other systems use it.
Final thought, because the popular LLMs don’t necessarily use it. Bing and Google do so; as long as Google and Bing are useful to your business, you should use it.
AIO and Inaccurate Information
There were a few questions on hallucinations (made-up information) and incorrect information presented in AI Overviews. One specific question from an attendee was about incorrect company information, and specifically phone numbers. She then asked why/if the business validated information from Google Business Profiles could be leveraged. There was no specific answer from the panel. They did state that those are two different systems, and they are working on ways to improve answer accuracy.
The attendee indicated she and others have posted on social media to the Search Liason, Danny Sullivan, and for others at Google with no response. Danny suggested that for more than just social posts, click the thumbs down button for the result and explain why. The teams can act better as these negative experiences increase via the feedback module.
We agree with Lilly Ray’s response in her recap. She pointed out that they are impacting businesses and that these sorts of results and lack of verification with their systems ” contradict Google’s emphasis on E-E-A-T and presenting trustworthy content.”
This is why businesses must understand what is presented not only in Google AIO results but also by other AI agents. This is another reason we have advocated for companies to adopt an Answers Management program. It is not precisely easy to change this information, but knowing it is wrong, identifying the source, and trying to correct it, and hammering that feedback button to try to elevate this as a bug.
Search is Evolving and We Must Evolve
After AIO, the second most significant topic was “evolution.” Multiple references were made to evolving and moving away from constants.
Multiple speakers reiterated this idea of evolution and listed the following five reasons that are driving it and why we must change with them. This seems to be where all the different teams align, trying to create a better product based on their data and insights. As Mike King asked, please share some of them so we can educate clients on this shift.
I don’t recall if a Google person said it or in one of our conversations, but like anything, you need to follow the money. Just remember, Google is a for-profit organization and is offering engaging and valuable services that enable it to make money. Point 1 in the screen capture states that users are “turning to closed-garden platforms for information needs previously met by search engines,” and the other two points signify change and the need to adapt for both customers and profit.
Item 3, Solving for Multiple Users and Behavior Patterns, indicated that things are changing, and we must evolve. We need to understand the SERPS results page and the searcher’s intent and what is best. Where applicable, we must ensure relevant coverage in any of the content blocks presented.
Reputation Abuse, Scaled Content
Danny spent a lot of time on updates, spam, and site reputation abuse manual actions, indicating two key conditions under which you will get hit for reputation abuse. One of the key changes to their guidelines what that sites that are “using third-party content on a site in an attempt to exploit the site’s ranking signals is a violation of this policy.” They gave these suggestions for removing the penalty. Since this content is often created by freelancers he wanted to emphasize that using freelancers is not the problem, it is how you use that content they create.
Google is a Collection of Systems
In his intro John Mueller gave an overview of how Google works. He mentioned they are a collection of systems that in aggregate give a result. We have talked about this on previous episodes related to their Search Off the Record podcast on how these systems may not work together as well as they should.
John emphasized the crawling team and the challenges they face with new technologies but said that the “decisions on what to present in the SERPS rests with the “serving team.” We hear a similar reference in the episode on duplicate content handling. If you think about serving, they are trying to match not only the best result based on the server’s intent but also present the right type of content and find ways to generate revenue for Google.
AI-Generated Image Labeling
Google stated that they are getting more strict on the need to label images as AI created and this is especially true for Shopping images. I have seen entries in the International Search Awards where agencies are creating these images at scale and are unsure if they are tagging them.
Alignment across Data Tools
This was an interesting chart showing what the different tools, Google Analytics, Trends, and Search Console, collect and what could be potentially connected. Daniel went on to explain the issues with date and time ranges and the depth of capture, making them nearly impossible to fully integrate.
