Last week was a good week to be an SEO in Boston.

As someone who lives in the area, it’s not often that major search events come to your backyard. Between the SEOFOMO meetup hosted by Aleyda Solís and WhitePress and three days of SMX Advanced, the search industry effectively took over Boston for a few days. Beyond the presentations, it was an opportunity to reconnect with friends, finally meet people I’ve interacted with online for years, and make a number of new connections along the way.

Having SEOFOMO immediately followed by SMX Advanced created something fairly unique. It gave attendees the opportunity to hear dozens of perspectives on the future of search over the course of a single week.

As I reviewed my notes afterward, I found myself comparing them to my notes from SMX Advanced 2025. Initially, I expected to find more discussions about AI, more conversations about agents, and more technical sessions about retrieval systems and large language models. Those things were certainly present. What surprised me, however, was how much the underlying conversational themes had changed in just one year.

The industry is still talking about AI. In fact, AI was the dominant topic across much of the week. The difference is that last year, we were largely discussing the consequences of AI. This year, we were discussing how these systems actually work and what that means for visibility moving forward.

Last Year: The Industry Was Focused on Traffic Loss

Looking back at my notes from SMX Advanced 2025, the dominant theme was disruption. AI Overviews were expanding. Publishers were reporting declining traffic. Click-through rates were falling. New AI-powered search experiences were appearing faster than most organizations could adapt to them. Many of the sessions focused on understanding what these changes meant for search marketers and website owners.

Michael King discussed the reality that some traffic loss was likely permanent. Chris Sullivan presented data around AI Overviews and their impact on clicks. Dave Davies explored the implications of agentic systems. Crystal Carter discussed reasoning models and how AI-generated answers were changing the search experience. Across multiple sessions, the conversation repeatedly returned to the same concern: if users receive answers directly from search engines and AI assistants, what happens to the traditional flow of traffic to websites?

The underlying question? How do we adapt to a world where AI is taking traffic?

That question shaped much of the discussion in 2025. Rankings, clicks, and website visits had long served as the primary measures of SEO success. When those metrics began changing, the industry naturally focused on understanding the impact.

This year moved in a different direction.

This Year: Nobody Was Arguing About AI Anymore

One of the most striking observations from both SEOFOMO and SMX Advanced 2026 was how little time was spent debating whether AI mattered. That conversation appears to be over.

Perhaps the most repeated idea throughout the week wasn’t a technical concept at all.

It was brand.

Speakers were no longer trying to convince audiences that AI would change search. There weren’t any presentations that I saw focused on proving that AI-powered search experiences were important. Instead, most speakers operated from the assumption that AI-assisted search, AI Overviews, AI Mode, ChatGPT, Gemini, and other recommendation systems are now part of the search landscape.

The focus went from “Will AI change search?” to “How do these systems retrieve, understand, and recommend information?”

Instead of focusing on what AI is doing to traffic, the industry has now started focusing on how AI systems evaluate information, establish trust, retrieve supporting evidence, and determine which sources deserve to be surfaced. That shift appeared repeatedly throughout the week and manifested in several themes.

Theme #1: Visibility Is No Longer the Goal

One of the most consistent themes across the week was that visibility alone is an incomplete measure of success.

Aleyda Solís discussed measuring AI presence, readiness, and business impact rather than focusing exclusively on traffic. Purna Virji challenged marketers to stop relying on time-saved metrics when discussing AI initiatives and instead connect those efforts to business outcomes. James Wirth explored the challenges associated with tracking AI visibility when answers, citations, recommendations, and retrieval paths can change constantly. Kelsey Libert shared research showing how trust and discovery are evolving as consumers engage with an increasingly fragmented set of platforms.

The common thread running through these discussions was that influence increasingly occurs before a click.

Users may encounter a brand in an AI Overview, an AI assistant, a recommendation engine, or an AI-generated answer without ever visiting the website. That interaction can still influence awareness, trust, consideration, and purchasing decisions.

Traffic still matters. However, traffic alone no longer tells the complete story.

Theme #2: Query Fan-Out Fully Entered the Conversation

One topic that repeatedly surfaced throughout the week was query fan-out. I’ll allow myself a small victory lap here because it’s something I’ve been writing about long before most people were.

This is something I first called out two years ago in my article, Google AI Overviews: Do Ranking Studies Tell the Whole Story?, before the industry had widely adopted the idea of what came to be known as “query fan-out.” My article pointed out how this was misguided. Here’s my conception from that article of how AI Overviews are built:

At the time, AI Overview research ad discussions focused on direct-match queries. However, after my earlier review of Google’s AI Overview patent, I noted that the retrieval process extended beyond the user’s original query to include related queries, recent queries, implied queries, and even location-based context. In other words, AI Overviews weren’t necessarily selecting sources based on a single query and retrieving responsive documents in a linear fashion. They were potentially gathering information through a much broader retrieval process. AI mode works in a very similar fashion according to its own patent.

That observation eventually to the first study (in 2024) of query fan-out with Authoritas examining whether related and reformulated queries could help explain AI Overview source selection. A note here that we referred to this phenomenon as “related queries” as “query fan-out” wasn’t established as the preferred term as of yet. We found that looking beyond the original query significantly increased the percentage of AI Overview links that could be accounted for in Google’s search results, suggesting that source selection was influenced by a broader set of retrieval paths than many ranking studies were considering at the time. I strongly recommend reading the 3 articles linked to get a good idea of how AI Overview and AI Mode works and query fan out.

This year, concepts related to query fan-out were discussed repeatedly throughout the conference. Dave Davies discussed retrieval behavior and how AI systems gather supporting information before constructing answers. Dawn Anderson’s keynote on RAG, context graphs, and retrieval systems naturally connected to the same idea. Will Scott’s presentation on evidence, entities, and AI visibility touched on many of the same underlying principles. Even discussions around AI citations and measurement pointed toward a broader reality: visibility increasingly depends on participation in retrieval systems rather than simply ranking for a single query.

In many ways, query fan-out helps explain why rankings alone no longer provide a complete picture of visibility. A business may not rank prominently for the original query and still become part of the final answer because it appeared somewhere within the retrieval process. Conversely, a business may rank well yet fail to appear in the retrieval paths that ultimately shape the recommendation.

Theme #3: Evidence Is Becoming the New Ranking Signal

Another theme that followed me throughout the week was evidence. Earlier in the week, Darren Shaw and Celeste Gonzalez, in a discussion of my Search Engine Land article, Google Ask Maps: How to optimize for visibility, discussed the idea that local SEO is becoming “evidence optimization.” What struck me was how often speakers discussed concepts that all pointed back to the same underlying idea.

Beth Nunnington spoke about earned proof and the importance of a brand’s digital footprint. Chris Sullivan discussed the growing role of digital PR and authentic mentions. Will Scott focused on evidence across source types and how machines develop confidence in their understanding of an entity. Conversations around reviews, Reddit, social proof, citations, community discussions, expert commentary, and media coverage appeared repeatedly throughout SEOFOMO and SMX.

Viewed individually, these can look like separate tactics. Viewed collectively, they begin to resemble something much larger.

Evidence. And, what about evidence?

• Evidence that a business exists.
• Evidence that people trust it.
• Evidence that it serves customers successfully.
• Evidence that others are willing to talk about it, reference it, and recommend it.

For years, some SEO conversations often centered around what one could publish about themselves. Other’s ahead of the game knew SEO expanded beyond owned channels and worked in parallel with content promotion, brand, reputation, and digiatal PR. They are now in the best position to succeed in the AI landscape.

Increasingly, AI systems are evaluating what everyone else says about us as well. That is significant because evidence is much harder to manufacture than content. Reviews have to be earned. Mentions require relationships. Community participation requires actual engagement. Expertise requires demonstrating knowledge over time.

As AI systems become more dependent on retrieval, citations, and corroborating information across multiple sources, evidence becomes increasingly difficult to separate from visibility.

Theme #4: Entity Understanding Is Becoming the Foundation

One of the biggest differences between the conversations in 2025 and 2026 was the prominence of entities, context, and relationships. Entity discussions are certainly not new. Google has been moving toward entity-based understanding for years. Knowledge graphs, schema, and entity relationships have been part of SEO conversations for more than a decade.

What was different this year was the degree to which these concepts moved from being technical SEO topics to becoming central to discussions about AI visibility. Dawn Anderson discussed context graphs and retrieval systems. Martha van Berkel explored enterprise knowledge graphs and how organizations can better control the information AI systems use to understand their brands. Grant Simmons discussed entity optimization and mapping meaning rather than simply targeting keywords. Dave Davies connected retrieval systems and citations back to how machines interpret information.

Across those discussions, a common theme emerged.

• Keywords alone don’t provide enough context.
• A mention without understanding isn’t particularly useful.

Even accurate information can be difficult for machines to use if they cannot connect it to a clear entity, topic, expertise area, category, or relationship. The challenge is no longer just helping search engines find information. The challenge is helping systems understand holistically what that information means and how it all conntects.

When viewed through that lens, concepts like schema, entity linking, knowledge graphs, and structured data become about creating clarity. The better machines understand who you are, what you do, and how you’re connected to other entities, the easier it becomes for them to confidently represent and recommend you.

Theme #5: Retrieval Matters More Than Ranking

If there was one technical topic that repeatedly surfaced throughout the week, it was retrieval. Retrieval became the bridge connecting nearly every major theme discussed during SEOFOMO and SMX. Whether speakers were discussing AI citations, grounding, query fan-out, RAG, entities, or brand visibility, the conversation often led back to the same underlying question: how does an AI system decide what information to retrieve before generating an answer?

Dawn Anderson’s keynote focused heavily on RAG, context graphs, retrieval systems, and the architecture that supports modern AI experiences. One observation that stood out was her statement that many AI failures are actually retrieval failures. If a system cannot retrieve the right information, it cannot generate the right answer. While AI systems often receive the blame when answers are incomplete or inaccurate, the problem frequently starts earlier in the process. The system can only work with the information it is able to find, retrieve, and connect.

Sam Torres highlighted another challenge: AI systems don’t always see the web the same way users or search engines do. Because crawling, rendering, and indexing are separate processes, a page can perform well in traditional search while remaining difficult for AI systems to access or understand. As AI retrieval becomes more important, how content is delivered matters just as much as where it ranks.

That same idea appeared repeatedly throughout the week. Dave Davies discussed retrieval signals and their relationship to AI citations. Will Scott focused on managing evidence across multiple source types so machines can better understand brands and entities. He also cautioned the audience against assuming that strong rankings automatically translate into AI visibility. Just because a page ranks well does not mean it will be retrieved, selected, or cited when an AI system generates a response. Conversations about grounding, AI visibility, citations, and query fan-out all pointed toward the same reality.

For years, ranking served as the primary proxy for visibility. If you ranked well, you were visible. If you didn’t rank, you weren’t. AI-powered search systems complicate that model. A page can rank highly and still fail to influence an answer. At the same time, a source can be retrieved, cited, and incorporated into a response despite never appearing prominently in traditional rankings.

The implication is that ranking is becoming an input rather than the finish line. Visibility is no longer determined solely by where a page appears in search results. It is increasingly determined by whether information can be retrieved, understood, trusted, and incorporated into the answer generation process.

The goal is becoming part of the retrieval process itself.

Theme #6: Brands Need Consistent Meaning

Perhaps the most repeated idea throughout the week wasn’t a technical concept at all. It was brand.

The word appeared in different contexts throughout SEOFOMO and SMX. Sometimes speakers discussed reputation. Sometimes they discussed trust. Sometimes they discussed authority, expertise, PR, entity understanding, or consistency. While the wording may have varied, they were often describing the same underlying challenge: helping both people and machines develop a clear understanding of who a brand is and what it represents.

This theme surfaced throughout the week and was reinforced during the closing Ask the Experts session featuring Martha van Berkel, Michael King, and Duane Forrester. By that point, after several days of discussions about retrieval systems, entities, citations, schema, AI visibility, and brand discovery, many of the conversations seemed to converge around a simple idea: brands that are easy to understand are easier to recommend.

Beth Nunnington spoke about how consumers experience brands in fragments across multiple platforms and touchpoints. Grant Simmons discussed defining entity relationships and the core topics a brand should be associated with. Martha van Berkel focused on ensuring AI systems receive accurate and connected information about organizations. Michael King emphasized understanding how brands are represented across the broader digital ecosystem and how that representation influences visibility.

Collectively, these discussions pointed toward an obvious observation: strong brands create consistent meaning. They consistently communicate who they are, what they do, who they help, and what they stand for. That consistency allows both people and machines to develop confidence in their understanding of the brand.

The opposite is also true. Conflicting information creates confusion. Inconsistent messaging creates uncertainty. Missing information creates gaps that machines are often forced to fill on their own. As search continues evolving into a collection of retrieval systems, recommendation engines, AI assistants, and answer platforms, consistency becomes increasingly important because systems cannot confidently recommend what they cannot confidently understand.

This theme tied together everything else discussed throughout the week. Query fan-out, evidence optimization, entity understanding, retrieval systems, citations, and AI visibility all ultimately support the same objective: creating enough clarity and confidence that both humans and machines understand who you are and why you matter.

The Movement From Visibility to Understanding

When I compare my notes from SMX Advanced 2025 to the conversations that took place during SEOFOMO and SMX Advanced 2026, one distinction stands out above everything else. Last year, much of the discussion centered around visibility. This year, the conversation focused on understanding.

From AI’s Impact to AI’s Mechanics

That doesn’t mean visibility has become unimportant. Rankings, traffic, clicks, impressions, and citations still matter because businesses still need to be discovered. What changed was the industry’s focus. In 2025, many of the conversations were centered on the consequences of AI. Speakers discussed declining click-through rates, AI Overviews, changing search behavior, answer engines, and what those developments might mean for publishers, businesses, and SEO professionals. The primary concern was understanding the impact AI would have on search and determining how organizations should respond.

This year, the discussion spent much less time on the consequences and much more time on the systems themselves. Retrieval, query fan-out, grounding, citations, evidence, entities, knowledge graphs, and trust appeared repeatedly throughout the week. Rather than asking whether AI would change search, speakers were examining how AI-powered systems gather information, evaluate sources, connect concepts, and determine what deserves to be included in an answer or recommendation. The conversation became less about what AI is doing and more about how AI works.

Understanding Becomes the Optimization Challenge

That distinction matters because it changes how we think about optimization. If visibility is primarily a ranking problem, then the goal is to improve rankings. If visibility is increasingly tied to understanding, then the challenge becomes much broader. Search engines and AI systems need enough information to understand who you are, what you do, who you help, and why you are credible. They need evidence that supports your expertise, context that connects you to relevant topics and categories, and corroboration from sources beyond your own website.

Many of the themes discussed throughout the week connect back to this idea. Query fan-out helps explain how information is gathered. Retrieval determines what information is considered. Evidence helps establish credibility. Entity understanding provides context. Brand consistency creates confidence. While these concepts are often discussed independently, they are all contributing to the same objective: helping machines develop an accurate understanding of a business before deciding whether it deserves to be recommended.

The New Question

That may be the biggest observation I took away from the week. Last year, the industry was largely focused on how AI would affect visibility. This year, the focus was on how AI develops understanding. The challenge now is moving beyond just getting discovered. Now the challenge is ensuring that when these systems evaluate the available evidence, they can confidently understand who you are and why you belong in the answer.

Final Thoughts: The Four Questions Every Brand Must Answer

As I reviewed my notes after the week ended, I realized that many of the presentations were ultimately addressing the same challenge from different perspectives. Some speakers approached it through retrieval systems and RAG. Others focused on query fan-out, evidence optimization, AI citations, schema, knowledge graphs, digital PR, brand building, or entity understanding. While the terminology varied, the underlying objective remained remarkably consistent.

At its core, modern search appears to be becoming a process of understanding. Before a search engine, AI assistant, or recommendation system can confidently recommend a business, it first needs to develop confidence in what that business represents. It needs enough information to connect facts, enough evidence to establish credibility, and enough context to understand where that business fits within a broader topic, category, industry, or local market.

When viewed through that lens, many of the discussions throughout SEOFOMO and SMX seemed to converge around four fundamental questions.

• What are you?
• What do you do?
• Who do you help?
• Why should someone trust you?

Google Search, AI Overviews, AI Mode, ChatGPT, Gemini, Ask Maps, and future agentic systems may all operate differently. They use different retrieval systems, different ranking methods, different interfaces, and different ways of presenting information. Yet they appear to share a common requirement. They need enough information, evidence, context, and confidence to understand what a business represents before they can recommend it.

That may be the biggest takeaway I brought home from last week. Last year, much of the industry’s attention was focused on traffic loss and the impact AI might have on search. This year, the conversation focused on understanding how these systems retrieve information, establish trust, and determine what deserves to be included in the answer. The questions have become less about what AI is taking away and more about what businesses need to do to be understood.

If the conversations from SEOFOMO and SMX Advanced are any indication, that discussion is only getting started.