The Oregon Convention Center in Portland is not the kind of venue where historic statements tend to be made. But on the morning of August 28, 2025, a room full of WordPress developers, bloggers, and content creators heard something that cut through nearly a year of industry anxiety in about thirty seconds.
Danny Sullivan — a man who founded Search Engine Watch, coined the phrase "search engine marketing," and spent seven years as Google's public-facing voice on search — stepped onto the stage at WordCamp US for what many in the room understood was his final major public appearance in his long-held role. He had stepped down as Google's Search Liaison on August 1, 2025, transitioning to an internal advisory role. The SEO community had been waiting to hear what he would say.
What he said, stripped of the trademark self-deprecating humour, was this: every new acronym the industry invents to describe optimising for AI search — GEO, AEO, AIO, LLM SEO — describes the same core practice SEOs have been doing for decades. Write content that genuinely helps people. Make it accessible, clear, and unique. That is it. That is the whole thing.
The message was both reassuring and, for a corner of the industry that had been selling "GEO audits" and "AI optimisation frameworks" for the better part of a year, more than a little pointed.
Who Is Danny Sullivan — and Why Does His Opinion Matter?
Danny Sullivan is, depending on your vintage in the industry, either a legend or simply the person who defined how the SEO world talks about itself. He founded Search Engine Watch in 1996, a site that became the primary source of record for search engine developments at a time when most people had barely heard of Google. He later founded Search Engine Land. He coined the term "search engine marketing." For two decades, he was the industry's most authoritative outside voice.
In 2017, he joined Google as its Public Liaison for Search — a role that involved explaining how Google's systems work to webmasters, journalists, and the broader SEO community, and feeding publisher concerns back into the company. He was, in effect, the translator between the opaque inner workings of Google Search and the people trying to make their living in it.
When Sullivan speaks about what Google actually values, it carries weight that most other Google employees' public statements do not. He built his credibility across three decades of explaining search to outsiders, and he spent seven years understanding it from the inside. His WordCamp keynote was not a press release or a structured Googlerspeech. It was a room full of people who make websites, asking him honest questions, and him giving honest answers.
WordCamp US 2025 — Oregon Convention Center, Portland
Sullivan delivered the keynote address at 9:15 AM PDT as the conference's headline speaker — his first major public appearance since stepping down as Google's Public Liaison for Search on August 1, 2025. The presentation covered the evolution of search, AI Overviews, E-E-A-T, and GEO. A Q&A at the end included a notable live exchange with blogger Angie Drake about declining click-through rates.
What Sullivan Actually Said — the Full Quote in Context
The way the phrase "Good SEO is good GEO" spread across the SEO industry in the days after the keynote stripped it of the context that made it land in the room. Here is what Sullivan actually said, verbatim, at approximately the 29:58 mark of his presentation:
"If you don't know what GEO is, it's like the latest acronym, but I can't keep track each day — there's a different one. But SEO, search engine optimization; GEO, generative engine optimization. By the way, if you could dig it out when I was like in 2010, back when people were panicking then, I was like, you know, SEO doesn't mean you get into the blue links on Google. SEO means you understand how people search for content and then you understand how to have your content there. And it could be everything from people asking a question to a voice device to people just opening up something on their phone or whatever. So, the basic things have not changed. Good SEO is good GEO, or AEO, AIO, LLM SEO, or LMNOPO. So, they're all fine. What I'm trying to say is don't panic. What you've been doing for search engines generally, and you may have thought of as SEO, is still perfectly fine and is still the things that you should be doing. Good SEO is really having good content for people."
Danny Sullivan, Director, Google Search — WordCamp US 2025, Portland, Oregon
The "LMNOPO" at the end was a joke — his way of mocking the industry's tendency to generate new acronyms for every slight variation of the same underlying practice. The laughter in the room was real and recognition was immediate. SEO professionals had spent the better part of twelve months attending webinars about GEO, buying consulting packages around AEO, debating whether LLM SEO required an entirely new technical stack. Sullivan's message was that most of that anxiety was noise.
→
→
→
→
→
Sullivan was not dismissing the fact that AI is changing how search results are displayed. He was saying that the content signals Google's systems use to decide what to surface in AI Overviews, in traditional results, and in generative features are the same signals they have always used. Unique, helpful content written for people. Strong page experience. Clear structure. These are not new requirements for a new era — they are the same requirements that have always defined what Google calls "good" content.
Sullivan Was Not Alone — Gary Illyes Said the Same Thing Weeks Earlier
Sullivan's WordCamp statement did not emerge in isolation. It echoed almost word-for-word a statement made by Google's Gary Illyes at the Search Central Live Deep Dive Asia Pacific 2025 event in Bangkok in July — roughly six weeks before WordCamp. The consistency between the two is notable because Sullivan and Illyes occupy very different roles at Google, and their public statements do not always align this closely.
Illyes went further on the technical side at the Asia Pacific event, explaining that from an engineering perspective there is no separate system for AI-driven results: the signals that feed into traditional ranking systems are the same signals that inform AI Overviews. He stated plainly that Google's algorithms train on the highest-quality content in the index, which is human-created, and that this fact had not changed with the introduction of AI features. Two senior Googlers, at two separate events, in two different months, saying the same thing. That is a coordinated message, not a coincidence.
The Moment That Made the Room Uncomfortable: The Angie Drake Exchange
The most discussed moment from the WordCamp keynote was not Sullivan's GEO statement. It was the Q&A that followed, and specifically the exchange that began at the 45:06 mark of the recorded presentation.
Angie Drake, who runs a responsible travel blog at notyouraverageamerican.com, stood up and asked Sullivan the question that thousands of independent publishers had been wanting to ask Google for months. Her click-through rate from Google search, she said, had collapsed since AI Overviews launched. She wanted to know what Google planned to do to compensate publishers who were losing traffic because users were getting the information they needed directly from AI-generated summaries — without ever clicking through to the source.
Angie Drake: "My click-through rate has tanked since you guys have started doing AI Overviews. I want to know what you're going to do to compensate bloggers who don't get clicks on their website because people are getting the information they need in the AI Overviews. I love AI. I love Google — but my click-through rate is just gone."
Danny Sullivan: "Well, unfortunately that's all the time we have. [Audience laughter] I'm not gonna say it at all. Can I just ask — are you seeing a rise in impressions and a drop in clicks?"
Angie Drake: "Yes. I've seen a rise in impressions. We're getting much more visitors. But our click-through rate is down. And click-through rate is something that sponsors are looking at."
Live exchange, WordCamp US 2025 Q&A — reported verbatim by multiple attendees
Sullivan's "Well, unfortunately that's all the time we have" — delivered before immediately continuing — got a laugh because the audience understood the instinct. He was not trying to dodge; he acknowledged that immediately and addressed the substance. His answer was careful: Google is unapologetic about providing direct answers to factual queries. Zero-click answers for questions like "what time is the Super Bowl?" have existed for years. The transition to AI-generated summaries for more complex queries is, in his framing, an extension of the same principle — serving the user directly rather than routing them through a publisher as an intermediary.
He acknowledged there would be "bumps along the way," that feedback was being heard within Google, and that solutions to the publisher traffic problem were still being worked through. He did not offer a timeline or a concrete commitment. That, for many publishers in the room, was the uncomfortable part of an otherwise reassuring presentation.
Sullivan's "Good SEO is good GEO" message is credible and well-supported by Google's public statements. But it does not resolve the legitimate concern Drake raised: if AI Overviews answer the user's query without sending traffic to the publisher, high-quality content still loses its economic value to its creator — even if it gets cited as a source in the AI summary. Being cited as a source in an AI Overview and receiving a click are not the same thing. Sullivan acknowledged this gap exists. He did not have a clean answer for it. Neither does anyone else yet.
What the Traffic Data Actually Shows for Publishers
The Impression-Click Decoupling Problem — and What It Means in Practice
The pattern Angie Drake described at WordCamp has been documented across multiple independent publisher analyses in 2025. Google's own internal data, referenced by Sullivan at the keynote, showed that AI Overviews drove a 10% increase in searches in the U.S. and India — users are running more queries, often longer and more complex ones. That part of the story is positive for search volume overall.
The complicating factor is the decoupling Sullivan asked Drake about: impressions rising while clicks fall. Research from Seer Interactive in mid-2025 found that the presence of an AI Overview in search results can reduce organic click-through rates by anywhere from 18% to 64%, depending on the query type. Informational queries — the kind that travel bloggers, recipe sites, and how-to publishers have built their businesses on — are most exposed, because those are exactly the queries AI Overviews are best at answering without requiring a click.
Sullivan's own data point — that AI Overviews have increased total search volume by 10% — represents the net effect across all queries. For publishers in informational categories, the shift is not net-positive. The growth in search volume is occurring in query types where AI Overviews are most likely to answer the question directly. For publishers with strong E-E-A-T signals, site authority, and content depth, the picture is more nuanced: some of these sites see referral traffic grow from AI Overview citations even as their keyword-by-keyword CTR falls.
What "Good SEO is Good GEO" Looks Like in Practice
Sullivan's message was broad. The practical question is what it means for someone sitting down to plan their content strategy for the next quarter. Sullivan, Illyes, and a series of supporting statements from Google's John Mueller point to a consistent set of principles — none of which are new, but several of which require more disciplined execution than they did before AI features became prominent in search results.
Write for People First
Sullivan's clearest statement: "Are you saying write things in a clear way that people can understand? That's just for people. Are you saying write about things that are unique or interesting? That's good for people." The test is not whether an algorithm rewards the content — it is whether a real person finds it genuinely useful.
The original standard
Uniqueness Over Repetition
Sullivan explicitly flagged the August 2025 spam rollout, which targeted low-quality AI-generated content. Content that repeats what is already available online — even well-written repetition — does not meet the uniqueness bar. Original angles, proprietary data, first-hand experience, and non-commodity perspectives are what Google is consistently rewarding.
Original or invisible
Structure for Readability and AI
Sullivan recommended clear headings, multimedia, and authoritative sourcing. Gary Illyes noted that AI bots from emerging platforms struggle with JavaScript-rendered content that Google's bots handle fine. Plain HTML structure with clear semantic headings performs better across both traditional and AI-driven retrieval.
Both humans and bots
E-E-A-T — Real, Not Performative
Sullivan warned directly against adding "expert reviewed" labels as a signal without genuine expert involvement. "What counts is actual expert knowledge in the content itself." E-E-A-T signals that are cosmetic — bylines without credentials, review badges without reviewers — are exactly the kind of pattern Google's quality systems are built to detect.
Real expertise only
Page Experience Still Matters
Sullivan included great page experience as a core element of both SEO and GEO. Mobile-first design, fast load times, accessible structure, and clear navigation to the "main thing" on each page. These signals feed into the same ranking layers whether results are presented as blue links or AI-generated summaries.
Technical hygiene
Measure Outcomes, Not Just Traffic
Sullivan noted that fewer but more engaged visits may be the new traffic profile for many publishers. Tracking sign-ups, leads, conversions, and time-on-site alongside raw visit count gives a more honest picture of content performance in an environment where AI Overviews are answering some queries before the click.
Engagement over volume
The Industry Counterpoint: Is Google Oversimplifying?
Search Engine Land's editorial response to Sullivan's keynote was worth reading alongside the quote itself. Its editorial team noted that while Sullivan's framing is accurate at the level of core content signals, GEO represents something genuinely new as a practice — not because the underlying content principles have changed, but because the output format of search has changed significantly. The future of Google and conversational AI search is answers, not rankings, regardless of what Googlers say publicly today.
This is a legitimate pushback. When the output of a search query is an AI-generated paragraph rather than ten blue links, the metric of success changes. Ranking position 1 through 10 all delivered traffic. Being cited as a source in an AI Overview may deliver no traffic at all, or a fraction of what a top-three ranking historically would. Even if the content signals required to earn that citation are identical to traditional SEO signals, the business impact of that citation is materially different from a ranking.
A March 2026 analysis on the relationship between GEO and SEO foundations found strong evidence that AI visibility is largely downstream of traditional search authority — that models surfacing results in live retrieval experiences tend to favour content that already performs well against standard search signals. The same analysis found that when Google removed the num=100 search parameter in September 2025, Reddit's appearance in ChatGPT responses dropped from roughly 60% to around 10%. That single data point illustrates just how tightly AI citation behaviour is currently linked to traditional search infrastructure.
"The signals haven't changed. The scoreboard has. And that matters more than Google's public position suggests."
What Has Changed vs. What Hasn't: An Honest Assessment
| Area | What Sullivan Says Hasn't Changed | What Has Genuinely Changed |
|---|---|---|
| Content signals | Same Unique, helpful, people-first content | Display Content may be surfaced without a click |
| Technical foundation | Same Page speed, mobile, crawlability | New gap Some AI bots can't parse JavaScript |
| E-E-A-T requirements | Same Real expertise, real authority | Higher bar Cosmetic signals now penalised more aggressively |
| Success metric | Changed Rankings no longer equal traffic | New Citation vs. click-through is the new distinction |
| Query complexity | Same Answer the intent behind the query | Expanded AI fans out complex queries into sub-searches |
| Revenue model for publishers | Changed Impressions ≠ ad revenue | No answer Google has not resolved the CTR-revenue gap |
| Optimisation vocabulary | Same GEO = good SEO by another name | Confusing Acronym proliferation creates unnecessary anxiety |
Know Which Channels Actually Drive Results
Sullivan's message is that content fundamentals haven't changed. But where your traffic comes from is shifting. Trimrly short links tell you exactly which source, platform, or campaign sent each visitor — so your content strategy is built on data, not assumptions.
What This Means for Your Content Strategy Right Now
- ✓
Stop auditing your content for GEO and start auditing it for quality. The question Sullivan is asking is not "is this content structured for AI retrieval?" — it is "is this content genuinely better than everything else available on this topic?" If the answer is yes, the AI retrieval problem largely resolves itself. If the answer is no, no amount of GEO-specific optimisation will substitute for content depth.
- ✓
Use structured headings, clear sections, and direct answers. Gary Illyes noted at the Asia Pacific event that AI features benefit from the same structural cues that have always helped humans navigate content. Descriptive H2s and H3s that directly state what the section answers — rather than clever or vague headers — help both readers and AI systems extract value from the page.
- ✓
Invest in content no one else has. Sullivan specifically called out the August 2025 spam rollout, which targeted low-quality AI-generated material. The single clearest differentiator in a world where AI can generate adequate content on any topic is proprietary data, first-hand experience, original research, and genuine expert knowledge that cannot be replicated from publicly available sources.
- ✓
Measure conversion quality, not just traffic volume. Sullivan's own acknowledgement that top-line clicks may decline for some publishers while conversion rates improve for quality-optimised content is a signal to reframe success metrics. Visits that result in newsletter sign-ups, purchases, or bookmarked pages represent stronger engagement than high-volume shallow visits that bounce in seconds.
- ✕
Do not add cosmetic E-E-A-T signals without substance behind them. Sullivan was explicit: "expert reviewed" labels added for show do not help. What counts is actual expert knowledge demonstrably present in the content itself — through cited research, specific credentials disclosed in context, and demonstrably experience-based claims that could not have been written by someone without that experience.
- ✕
Do not chase word count targets or padding. Sullivan listed unnecessary word count targets as one of the myths around optimisation that publishers should actively ignore. Content should be as long as the topic requires to be fully useful to the reader — no longer, no shorter. Adding length to hit a perceived algorithm threshold is exactly the kind of behaviour Google's quality systems are designed to detect.
- ✕
Do not expect this to resolve the publisher revenue problem. Sullivan's message is reassuring about content strategy. It is not reassuring about business models built on advertising revenue tied to click volume. That problem remains real and unresolved, and building a content business that depends entirely on organic CTR from Google without diversifying audience acquisition channels introduces structural risk regardless of content quality.
Frequently Asked Questions
GEO refers to the practice of optimising content to appear as a cited source in AI-generated search responses — such as Google's AI Overviews, ChatGPT's web search results, or Perplexity's citations. It is sometimes also called AEO (Answer Engine Optimisation), AIO optimisation, or LLM SEO. Danny Sullivan's point at WordCamp US is that all of these describe the same underlying practice: creating high-quality, authoritative, clearly structured content that genuinely helps users. The display format of search results has changed; the content signals that determine what gets surfaced have not changed significantly.
At the level of content signals, largely yes — which is Sullivan's point. The same E-E-A-T principles, the same quality content requirements, and the same technical foundations that Google has recommended for traditional SEO are the signals that determine whether a page is cited in AI-generated results. Where GEO differs is in its success metrics: being cited in an AI Overview does not guarantee a click, and the business impact of an AI citation is materially different from a top-three organic ranking. The practice of creating the content may be the same; the downstream economics are not.
E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trustworthiness. It is the framework Google's quality raters use to assess content quality, and it has been a core part of Google's Search Quality Evaluator Guidelines for years. Sullivan's emphasis on it at WordCamp reflects its growing importance: as AI-generated content floods the web, Google's systems are increasingly focused on signals that distinguish genuinely expert, first-hand content from plausible-sounding but shallow material. Sullivan specifically warned against cosmetic E-E-A-T signals — labels and badges added for show without real expert involvement in the content itself.
The honest answer is: it depends on content type, and the picture is mixed. Sullivan's internal data showed a 10% increase in total searches in the US and India after AI Overviews launched, suggesting more query volume overall. But independent research found that when an AI Overview is present for a specific query, organic click-through rates can fall 18% to 64%. Informational publishers — travel blogs, how-to sites, recipe sites — face the most exposure, because those are the query types AI Overviews answer most effectively. Publishers with deep, unique content and strong authority signals tend to fare better than commodity-content sites. Sullivan acknowledged this problem exists and did not offer a resolved answer to it.
Danny Sullivan founded Search Engine Watch in 1996 and Search Engine Land, coined the term "search engine marketing," and spent decades as the SEO industry's most authoritative independent voice before joining Google in 2017 as its Public Liaison for Search. In that role he served as the translator between Google's internal search operations and the webmaster and SEO community, explaining how systems work and escalating publisher concerns internally. He stepped down from the Search Liaison role on August 1, 2025, transitioning to an internal advisory position. His WordCamp US keynote was widely viewed as a farewell address, and it carried the weight of someone who understands search from both sides of the wall.