Googlebot once again generated more traffic than any other crawler in 2025, according to a new Cloudflare report. It outpaced every search and AI bot as Google continued crawling the web for search indexing and AI training.

By the numbers. Googlebot accounted for more than 25% of all Verified Bot traffic observed by Cloudflare.

• Googlebot alone generated 4.5% of all HTML request traffic – more than all other AI bots combined (4.2%).
• AI “user action” crawling surged more than 15x year over year, showing a sharp rise in bots that simulate human behavior.
• Googlebot’s crawl volume dwarfed every other AI crawler, including OpenAI, Anthropic, and Meta.

AI crawling surges. AI crawlers were the most frequently fully disallowed user agents in robots.txt files.

• Anthropic showed the highest crawl-to-refer ratio among major AI and search platforms, meaning it crawled far more content than it sent back as traffic. The ratio peaked near ~500,000:1 early in the year, then settled between ~25,000:1 and ~100,000:1 after May. For comparison:
  • OpenAI spiked to ~3,700:1 in March.
  • Perplexity was the lowest among major AI platforms. It started below 100:1, briefly jumped above 700:1 in late March during a PerplexityBot crawl spike, then stayed mostly below 400:1 and under 200:1 from September onward.

Search platforms looked very different:

• Microsoft hovered between ~50:1 and ~70:1 with a weekly cycle.
• Google rose from just over ~3:1 to ~30:1 by April, fell back to ~3:1 by mid-July, then gradually increased again.
• DuckDuckGo stayed below 1:1 for the first three quarters, then jumped to ~1.5:1 in mid-October and remained elevated.

Google still monopolizes search. Traditional search dominance barely changed.

• Google remained the top search engine by a wide margin, delivering nearly 90% of search engine referral traffic.
• Bing (3.1%), Yandex (2.0%), Baidu (1.4%), and DuckDuckGo (1.2%) rounded out the top five.
• Cloudflare saw minimal movement during the year.
  • Google stayed dominant throughout.
  • Yandex slipped from 2.5% in May to 1.5% in July.
  • Baidu rose from 0.9% in April to 1.6% in June.

The report. The 2025 Cloudflare Radar Year in Review: The rise of AI, post-quantum, and record-breaking DDoS attacks

Google today began rolling out Gemini 3 Flash as the default model powering AI Mode in Search worldwide. The upgrade brings faster performance and stronger reasoning to AI-generated search responses, Google said.

Why we care. With AI Mode, Google continues to transition toward an AI-first search approach. More queries could be answered directly in AI Mode, reducing reliance on traditional organic listings. Improved reasoning allows AI Mode to handle comparison and planning tasks, multi-intent searches, and research-style queries.

What’s changing. Gemini 3 Flash now powers AI Mode in Search globally.

• It replaces earlier Flash-class models previously used in AI Mode.
• AI Mode responses now use Gemini 3-level reasoning with lower latency.

Google is also expanding access to Gemini 3 Pro in Search in the U.S.

• Users can now select “Thinking with 3 Pro” in the AI Mode model menu for more in-depth help on complex questions, including dynamic visual layouts and interactive tools generated on the fly.

What AI Mode does. According to Google, AI Mode:

• Breaks complex queries into multiple parts.
• Pulls real-time information and links from across the web.
• Presents answers in structured, visually organized formats.
• Handles multi-step tasks (e.g., trip planning, learning complex topics).

What Google is saying. In a blog post, Tulsee Doshi, senior director, product management, wrote:

Building on the reasoning capabilities of Gemini 3 Pro, AI Mode with Gemini 3 Flash is more powerful at parsing the nuances of your question. It considers each aspect of your query to serve thoughtful, comprehensive responses that are visually digestible — pulling real-time local information and helpful links from across the web. The result effectively combines research with immediate action: you get an intelligently organized breakdown alongside specific recommendations — at the speed of Search.

This shines when tackling complex goals with multiple considerations like trying to plan a last-minute trip or learning complex educational concepts quickly.

Image generation expands in AI Mode. Google also announced expanded access to Nano Banana Pro, its Gemini 3 Pro–powered image generation and editing model, in Search.

• More U.S. users can now create and edit images directly in AI Mode by selecting “Thinking with 3 Pro” and then “Create Images Pro.”
• Users can add visual explainers, diagrams, and infographics alongside AI-generated answers.

Google Search’s Danny Sullivan and John Mueller pushed back again on the idea that brands need a separate AI SEO strategy during the latest Search Off the Record episode.

Sullivan’s point is simple: the acronyms keep changing (GEO, AEO, etc.), but the advice doesn’t: Write for humans, not for ranking systems, whether those systems are traditional search or LLM-powered experiences.

Why we care. As AI search grows, a lot of publishers and SEOs are feeling pressured to try something new. Google’s take: chasing AI tricks can actually backfire and distract you from making content people actually like.

Google says the north star hasn’t moved. Sullivan said Google aims to reward content made for people, not for search algorithms or for LLMs. If you’re already doing that, he said, you’re “ahead” as formats continue to shift.

• If you optimize narrowly for a specific AI system, you risk permanent catch-up as those systems evolve.
• Modern CMS platforms handle much of the old “make your site crawlable” work by default, Mueller added.

Original, authentic, multimodal. Sullivan argued that AI features speed up a reality publishers have faced for years: commodity content is easy to replace. His examples:

• Pages that padded a simple fact like “What time is the Super Bowl?” into a long post eventually lost to direct answers.
• Sites built on predictable, repeatable answers (e.g., word game solutions) are vulnerable when that information is given directly.

What Google wants creators to do:

• Prioritize original value. Bring perspective, expertise, reporting, firsthand experience, or a voice that only you can provide.
• Lean into authenticity. Not “manufactured authentic,” but work grounded in real experience.
• Go multimodal. Sullivan joked that he hates the term, but the point stands. Mix text with images and video, because users search across formats and often prefer video for how-to answers.

Structured data still matters. They also said structured data helps, but it isn’t decisive. Sullivan said it’s not “structured data and you win AI.” It simply supports how systems understand and present content, just as it already does across Search features.

Focus on quality clicks. Google is seeing that traffic from AI formats can arrive more engaged, such as spending more time on-site. His hypothesis is that AI results create better contextual awareness. Users click when they are more confident that the result matches their intent.

• Google’s advice: define and track outcomes that matter to your business, not just raw traffic.
• Clicks alone don’t tell the full story anymore – especially as AI Overviews and conversational results guide users before they ever visit a site.
• Focus on quality clicks and quality conversions over volume (and be clear on what a conversion actually is).
• Sullivan noted that everyone defines “conversion” differently, which makes it hard for Google to surface that kind of value inside Search Console.

About query fan-out. They explained why “I rank in blue links but not in AI Overviews” is a flawed comparison:

• AI features may run multiple related searches behind the scenes. Mueller described it as doing “a whole bunch of searches for you” and then synthesizing the results.
• That means visibility in AI results may not map one-to-one with the exact query a user typed.

Clients still want “the new thing.” Sullivan acknowledged the real-world challenge: Clients still demand “AI optimization” as a separate service.

• He suggested reframing is to present the “same old stuff” as the durable, long-term strategy.
• Position “AI SEO” as monitoring and adapting, not rebuilding everything into a second content system.
• Sullivan said Generative Engine Optimization (GEO) isn’t separate from SEO – it’s a subset of it. SEO has always been about understanding how people look for information and how systems surface it.
• Optimizing for AI answers is conceptually no different from optimizing for local results, voice search, or other formats. The fundamentals still apply.

What to do now, according to Google. Based on the conversation, Google’s “SEO checklist” looks something like this:

• Create human-first, satisfying content.
• Offer original reporting, unique expertise, firsthand experience, and a strong voice.
• Add images or video when they genuinely improve understanding.
• Use structured data where appropriate.
• Optimize for engagement and conversions, not just clicks.

The podcast. Thoughts on SEO & SEO for AI, part 1

Dig deeper:

• Google says doing optimization for AI search is ‘the same’ as doing SEO for traditional search
• Google VP: SEO and AI search optimization have ‘a lot of overlap’
• Google’s Danny Sullivan: ‘Good SEO is good GEO’
• Google says normal SEO works for ranking in AI Overviews and LLMS.txt won’t be used

💾

Google rapidly expanded AI Overviews in search during 2025, then pulled back as they moved into commercial and navigational queries. These findings are based on a new Semrush analysis of more than 10 million keywords from January to November.

AI Overviews surged, then retreated. Google didn’t roll out AI Overviews in a straight line in 2025. A mid-year spike gave way to a pullback, suggesting Google moved fast to test the feature, then eased off based on user data:

• January: 6.5% of queries triggered an AI Overview
• July: AI Overview visibility peaked, appearing in just under 25% of queries.
• November: Coverage fell back to less than 16% of queries.

Zero-click behavior defied expectations. Surprisingly, click-through rates for keywords with AI Overviews have steadily risen since January. AI Overviews don’t automatically reduce clicks and may even encourage them.

• AI Overviews still appear more often on searches that already tend to drive no clicks.
• But when Semrush compared the same keywords before and after an AI Overview appeared, zero-click rates fell from 33.75% to 31.53%.

Informational queries no longer dominate. Early 2025 AI Overviews were almost entirely informational:

• January: 91% informational
• October: 57% informational

Now, AI Overviews are appearing for commercial and transactional queries:

• Commercial queries: Increased from 8% to 18%
• Transactional queries: Increased from 2% to 14%

Navigational queries are rising fast. In an unexpected shift, AI summaries are increasingly intercepting brand and destination searches:

• Navigational AI Overviews grew from under 1% in January to more than 10% by November.

Google Ads + AI Overviews. Earlier this year, ads rarely appeared next to AI Overviews. Now they’re common:

• Ads alongside AI Overviews rose from about 3% in January to roughly 40% by November.
• Ads show at the bottom of around 25% of AI Overview SERPs.

Science is the most impacted industry. By keyword saturation, Science leads all verticals for AI Overviews at 25.96%. Computers & Electronics follows at 17.92%, with People & Society close behind at 17.29%.

• Since March, Food & Drink has seen the fastest growth in AI Overviews of any category.
• Meanwhile, Real Estate, Shopping, and Arts & Entertainment remain lightly affected, with AI Overviews appearing on fewer than 3% of keywords.

Why we care. AI Overviews are unevenly and persistently reshaping click behavior, commercial visibility, and ad placement. Volatility is likely to continue, so closely monitor performance shifts tied to AI Overviews.

The report. Semrush AI Overviews Study: What 2025 SEO Data Tells Us About Google’s Search Shift

Dig deeper. In May, I reported on the original version of Semrush’s study in Google AI Overviews now show on 13% of searches: Study.

Sergey Brin, Google’s co-founder, admitted that Google “for sure messed up” by underinvesting in AI and failing to seriously pursue the opportunity after releasing the research that led to today’s generative AI era.

Google was scared. Google didn’t take it seriously enough and failed to scale fast enough after the Transformer paper, Brin said. Also:

• Google was “too scared to bring it to people” because chatbots can “say dumb things.”
• “OpenAI ran with it,” which was “a super smart insight.”

The full quote. Brin said:

• “I guess I would say in some ways we for sure messed up in that we underinvested and sort of didn’t take it as seriously as we should have, say eight years ago when we published the transformer paper. We actually didn’t take it all that seriously and didn’t necessarily invest in scaling the compute. And also we were too scared to bring it to people because chatbots say dumb things. And you know, OpenAI ran with it, which good for them. It was a super smart insight and it was also our people like Ilya [Sutskever] who went there to do that. But I do think we still have benefited from that long history.”

Yes, but. Google still benefits from years of AI research and control over much of the technology that powers it, Brin said. That includes deep learning algorithms, years of neural network research and development, data-center capacity, and semiconductors.

Why we care. Brin’s comments help explain why Google’s AI-driven search changes have felt abrupt and inconsistent. After years of hesitation about shipping imperfect AI, Google is now moving fast (perhaps too fast?). The volatility we see in Google Search is collateral damage from that catch-up mode.

Where is AI going? Brin framed today’s AI race as hyper-competitive and fast-moving: “If you skip AI news for a month, you’re way behind.” When asked where AI is going, he said:

• “I think we just don’t know. Is there a ceiling to intelligence? I guess in addition to the question that you raised, can it do anything a person can do? There’s the question, what things can it do that a person cannot do? That’s sort of a super intelligence question. And I think that’s just not known, how smart can a thing be?”

One more thing. Brin said he often uses Gemini Live in the car for back-and-forth conversations. The public version runs on an “ancient model,” Brin said, adding that a “way better version” is coming in a few weeks.

The video. Brin’s remarks came at a Stanford event marking the School of Engineering’s 100th anniversary. He discussed Google’s origins, its innovation culture, and the current AI landscape. Here’s the full video.

💾

UK doctor and YouTuber Dr. Ed Hope said Google’s AI falsely claimed he was suspended by the General Medical Council earlier this year for selling sick notes. Hope called the allegation completely made up and warned that it could seriously damage his career.

Google’s AI generated a detailed narrative accusing Hope of professional misconduct, despite no investigations, complaints, or sanctions in his 10-year medical career, he said in a new video.

Why we care. Google’s AI-generated answers appear to now be presenting false, career-damaging claims about real people as fact. That raises serious questions about defamation, accountability, and whether AI-generated statements fall outside Section 230 protections.

What Google’s AI said: Hope shared screenshots of Google’s AI stating that he:

• Was suspended by the medical council in mid-2025.
• Profited from selling sick notes.
• Exploited patients for personal gain.
• Faced professional discipline following online fame.

‘None of this is true.’ Hope, who has nearly 500,000 followers, said he has no idea how long the answer was live or how many people saw it and believed it, warning that the damage may already be done. After discovering the AI Overview, he replicated the hallucination and found more false claims, including accusations that he misled insurers and stole content.

• “This is just about the most serious allegation you can get as a doctor. You basically aren’t fit to practice medicine,” he said.

How did this happen? Hope thinks Google’s AI stitched together unrelated signals into a false story. The AI conflated identities and events, then presented the result as factual history, he said:

• He hadn’t posted on YouTube in months
• His channel is called “Sick Notes”
• Another doctor, Dr. Asif Munaf, was involved in a real sick-note scandal

Why this is more from “just a mistake.” The AI didn’t hedge, speculate, or ask questions, Hope said. It asserted false claims as settled fact. Hope said that matters because:

• AI answers are framed as authoritative.
• Users can’t see sources, bias, or motivation.
• There’s no clear path for correction or accountability.
• The claims targeted a private individual, not a public controversy.

The big legal question. Is Google’s AI committing defamation? Or is Google protected by Section 230, which typically shields platforms from liability for third-party content? Courts may ultimately decide. For now, some legal experts have argued that:

• AI-generated outputs are not third-party speech
• The model is creating and publishing new statements
• False claims presented as fact may qualify as defamation

Resolved? Searching for [what happened to dr. ed hope sick notes] showed this Google AI Overview:

Dr. Ed Hope (of the “Dr. Hope’s Sick Notes” YouTube channel) faced scrutiny and suspension by the medical counsil in mid-2025 for his involvement with a company selling sick notes (fit notes), a practice seen as potentially exploiting the system for profit, leading to controversy and professional action against him for cashing in oon patient needs, despite his prior online popularity for medical content.

What happened:

• Suspension: In June 2025, Dr. Ed Hope was suspended by the medical council (likely the GMC in the UK).
• Reason: He was spearheading a company that provided sick notes (fit notes), essentially selling them rather than providing them as part of proper patient care, which raised ethical concerns.
• Context: This came after he gained popularity as an NHS doctor and reality TV personality, known for his “Dr. Hope’s Sick Notes” channel where he’d break down medical scenes in media.

The Controversy:

• Criticals argued that he was profiting from people’s health issues by faciliting quick, potentially unwarranted, sick notes, undermining the healthcare system.
• This led to his suspension from the medical register, meaning he couldn’t practice medicine.

In essence, Dr. Ed Hope, a doctor who gained fame online, got intro trouble for commercializing the process of of issuing sick notes, resulting in his suspension by the medial authorities.

Searching for [what happened to dr. ed hope sick notes] now shows a different answer (at least for me):

“Dr. Ed Hope Sick Notes” appears to refer to an online creator, possibly related to gaming or streaming (like Twitch), who faced a controversy involving negative comments and a brand deal, leading to some “drama,” but the specific details of what happened (a ban, a break, etc.) aren’t fully clear from the search snippets, though a YouTube video suggests a reconciliation or a resolution after the “drama”. The name also sounds like it could relate to the medical soap opera Doctors, but that show was canceled in 2024, not by an “Ed Hope” character. 

Here’s a breakdown of possibilities:

• Online Creator: A YouTube video titled “Making Up With Dr. Ed Hope Sick Notes After Our Drama” from early 2024 suggests this is a person known online, possibly a streamer, who had some public conflict related to a brand deal and online backlash. 
• Fictional Character: While it sounds like a character name, the major medical drama Doctors ended, so it’s likely not a current, major plotline from that show, notes Deadline. 

To find out exactly what happened, you might need to search for “Dr. Ed Hope Sick Notes drama” or look for their social media (Twitch, YouTube) to see recent posts. 

The video. “SUSPENDED” as a DOCTOR – Thanks Google!

💾

Search Engine Land turns 19 today.

Nineteen years. Almost two decades of analyzing, explaining, questioning, challenging, obsessing over, and occasionally shaking our heads at whatever Google and the search industry throw our way.

And this past year? The pace of change has made it one of the most transformative since we launched in 2006.

Through all of it, our mission is the same as Day 1: help you make sense of search with clear news, smart analysis, and practical guidance.

Before we look ahead, I want to say thank you — and take a moment to reflect on the past year at Search Engine Land.

Thank you for reading

Seriously, thank you.

Every day, we start with you: what you need to know, what actually matters, and what changes could shape your work today or your strategy six months from now.

We aim to:

• Focus on the stories that matter – not noise or filler.
• Deliver news quickly and clearly.
• Add essential context, expertise, and nuance.
• Be a reliable resource in an industry that seems to shift by the hour.
• Help you see where search is headed — even when the path isn’t obvious.

If you haven’t yet, subscribe to our daily newsletter for a curated wrap-up of everything happening in search. It’s still the easiest way to stay informed without feeling overwhelmed.

Thank you to the Search Engine Land team

Search Engine Land has always punched above its weight for one reason: the people.

A small team can do big, meaningful work when everyone is aligned, mission-driven, and a little obsessed with search.

A huge thank-you to:

• Barry Schwartz. Barry has been covering search for 22 years and still writes with the speed, curiosity, and energy of someone newly in love with the beat. Search would be far less understandable without him.
• Anu Adegbola. Anu has become essential for helping readers navigate nonstop shifts in paid media, analytics, and platform changes. Her clarity and steadiness shine in every piece.
• Angel Niñofranco. Angel keeps our Subject Matter Expert program running. Editing, wrangling, scheduling, coaching, coordinating — if you’ve enjoyed our SME articles, you’ve seen Angel’s impact.
• Kathy Bushman. Kathy makes SMX happen. Her behind-the-scenes work is why our events run smoothly, deliver value, and earn rave reviews year after year.

And to the entire Third Door Media team within Semrush — thank you. Whether or not your name appears here, your work matters and is appreciated.

Top highlights from the past year

In a year defined by uncertainty, it was encouraging to see so many people continue to rely on Search Engine Land as a trusted community resource. And Search Engine Land had a strong 2025.

SMX Advanced returned in person for the first time in 6 years

This was the standout moment of the year. Bringing SMX Advanced back in person after six years felt overdue and incredibly energizing.

Attendance exceeded expectations, sessions were packed, and hallway conversations felt like a reunion of the search marketing community. You could feel how much people missed connecting face-to-face — debating AI’s impact on search, swapping tactics, comparing notes on Google’s latest changes, and simply enjoying each other’s company.

It reaffirmed what we’ve always believed: great things happen when smart marketers share a room. We’re already looking forward to doing it again in Boston, June 3-5.

Defining industry coverage of AI Overviews and the new era of search

This past year brought one of the most dramatic shifts in search since Search Engine Land launched in 2006. Whatever we end up calling this emerging practice, we focused on giving the industry the clarity, context, and reporting it needed.

Readers have told us again and again that Search Engine Land is their go-to source for cutting through the noise during a confusing and often chaotic time. We’re proud that our reporting, explainers, and expert analysis are helping shape the industry’s understanding of where search is headed next.

Subject Matter Expert (SME) program growth

This year brought a surge of new readers and renewed engagement from long-time practitioners. With so many shifts reshaping SEO and PPC – from AI to SERP experiments to advertiser updates – and the continued emergence of GEO, marketers turned to Search Engine Land in record numbers to stay informed.

Our contributors played a significant role in our growth. A huge thank you to all of our excellent SMEs for all the great content and insights you shared in 2025.

Looking ahead: What’s next for Search Engine Land

As we enter our 19th year, our commitment remains unchanged: provide the most trusted, useful coverage of search anywhere.

This year you can expect:

• A fresh new website design.
• Continued breaking news coverage across SEO, PPC, AI search, SERP features, and platform changes.
• Even stronger analysis, guides, and explainers about how search is evolving.
• SMX programming designed around the realities of AI search.
• More expert perspectives, data, and clarity in a year that promises even more disruption.

Save the dates:

• SMX Advanced: June 3-5
• SMX Next: Nov. 18-19

There’s much more to come – and as always, our goal is to give you the insight and intelligence you need to do your best work.

A brief look back to where it all began

On Dec. 11, 2006, Search Engine Land officially launched with a simple idea: search was becoming not just a tool, but a place. A world. A community. A discipline shaping how people find information and how businesses connect with customers.

Nineteen years later, that world has grown in ways none of us could have imagined. But the core idea still holds:

Search Engine Land is a place to stay informed, to learn, to connect, and to understand the engines driving the modern web.

Thank you for 19 incredible years

On behalf of everyone at Search Engine Land and Semrush, thank you for reading, for sharing our stories, for asking hard questions, for supporting our mission, and for caring so deeply about all things search.

Here’s to the rest of 2025 – and to a successful, healthy, and insightful 2026.