How Google’s AI Search is Reshaping SEO Strategy

How multimodal expectations and query fan-out are pushing industry norms.

Google’s evolution of search isn’t just about new tools it’s a reshaping of how information is discovered, indexed, and valued. With the rollout of AI Mode in Search, Google is unlocking capabilities powered by query fan-out and multimodal understanding — and as SEOs we need to rethink everything from keyword research to content creation in order to keep pace.

Query Fan-Out: The End of Single Keyword SEO

Traditional keyword research operated on the assumption that a single phrase or term could be optimized against a specific query. This has been a long time false reality. Since 2009 Google has worked hard to expand the understanding of query to intent and has built countless systems to refine results that match expectations, avoiding the linear concept of the keyword. Google’s new AI Mode expands this paradigm and unlocks more direct answers to users’ questions and reshapes the field of optimization for search. 

Traditional keyword research operated on the assumption that a single phrase could be optimized to match a specific query. But that’s been a flawed premise for over a decade. Since 2009, Google has steadily moved beyond literal keyword matching, building systems that interpret intent and refine results accordingly — rejecting the linear model of keyword-to-result.

This shift is backed by a long lineage of patents that reveal how Google has continuously advanced its understanding of language, personalization, and refinement. These innovations laid the groundwork for today’s AI-powered search capabilities:

2003: Search Query Refinement, US8645407B2 (Inventor: Paul Haahr)
2004: Autocompletion of Partial Search Query, US8515954B2
2009: Expanded Sibling Query Refinements, US8667003B1
2013: Speech Recognition to Query,  EP2909832B1
2013: Search Query Based on Personal Information, US9317585B2

These patents, among many others, illustrate Google’s long-term intent: to move from matching strings to understanding meaning.

What is Query Fan-Out?

With the introduction of AI Mode, this evolution takes a major leap forward. Google now deconstructs user queries into layered intents using query fan-out, delivering more direct, context-aware answers. As a result, the entire field of search optimization must adapt to a reality where relevance is determined by meaning, not just match.

As outlined in Google’s announcement on AI Mode, AI Mode now uses query fan-out to deconstruct a user’s search into subtopics and then issues multiple simultaneous queries to deeply explore each intent. This allows AI-powered search to tap into a broader and more relevant pool of content across the web. More importantly, it even allows AI mode results to target outcomes for users looking for specific outcomes in the results. 

The SEO community and content professionals must adapt. The new frontier is not keywords — it’s research concepts. Keyword research must evolve into topic and subtopic mapping, supported by a combination of both internal data and refinement of insights using techniques already defined by Google. 

These refinements are no longer distant realities; they are prompts that can refine our research and vectorizing to build our own knowledge graphs of data. There are hundreds of Google patents that outline how they are consuming, refining, and interpreting data to inform decisions that consumers via search. Here is a great list of patents covered in Search Engine Land’s 2023 patent summary. These elements are the foundation of technologies like BERT and BARD, which trained machines to understand semantic relationships and contextual meanings.

This shift demands that we stop thinking in siloed queries and start thinking like information architects, identifying the layered questions, intents, and informational needs within a given topic cluster.

Deep Research & the New Expectations of Content

The release of Deep Search within AI Mode takes query fan-out even further. Google can now issue hundreds of queries on a user’s behalf, aggregating and reasoning across scattered information to generate direct responses with fully cited answers.

In AI Mode, Google is going beyond searching relevance and connecting to the users expectations. This is the requirement that is still evolving, meeting a user expectation within the content and context developed for a user’s outcome, that is defined by a path toward synthesizing them into a multidimensional response.

This introduces a new bar for content quality. It’s no longer enough to answer a primary query—your content must support subtopic queries, user discovery, and layered information paths with a value experience that surfaces within AI responses. 

So how do you define research in this new paradigm?

Let’s take Office Space Software as a working example. The brand currently ranks for 155 keywords featured in AI Overviews. To analyze this performance, we vectorized the full keyword set to reveal how Google associates meaning across related topics. The resulting two-dimensional plot provides a visual map of semantic proximity and clustering. (See visualization below.)

One insight was the significant distinction between themes like “AI at work” versus “AI in the workplace.” The latter connected with far more meaningful commercial intent and content potential. Conversely, outliers such as “space in office” or “flexibility” were loosely associated yet likely to yield minimal success if pursued without depth or alignment to high-value queries. These insights informed a clear path for content investment based on AI Mode’s understanding — not guesswork or keyword stuffing.

Reaching this level of clarity in research means expanding beyond keyword tools. The most strategic research begins with internal knowledge assets:

• Customer feedback and surveys
• Support documentation and help desk tickets
• Internal sales scripts, call transcripts, and Q&A logs
• Sales and product performance data mapped to topic categories

These sources reveal gaps, opportunities, and language that no third-party tool can surface with precision. For SEO professionals looking to win in the era of AI-driven search, internal knowledge must become the foundation of content strategy.

This is what Google is training its systems to understand: not just what users search for, but what real-world expertise looks like when it’s made accessible.

We’ve already seen examples of this shift play out in competitive spaces. Take Office Space Software, a brand with over 155 keywords that currently rank with AI Overviews. By vectorizing these keywords, we quickly surfaced patterns about how Google’s system associates themes.

These findings demonstrate how SEOs must move from ranking for terms to aligning content with Google’s understanding of subtopic clusters. It also sets a new expectation for how research should be conducted.

The most critical — and often overlooked — component of modern SEO research is internal knowledge. Too often, SEOs rely solely on third-party tools or even Google Search Console. But the richest sources of insight are often already within your organization:

• Customer feedback and surveys
• Support tickets and documentation
• Sales scripts and call transcripts
• Product performance data mapped to queries or categories

These internal datasets can unlock nuanced content opportunities that align tightly with user needs—and with the way AI search systems reason and retrieve.

Content Optimization Is Not What It Used To Be

The SEO optimization concept as we once knew it is a worthless act of aggregated, misled data points. Adding a few keywords, rewriting meta descriptions, or awkwardly inserting CTAs no longer moves the needle. For years, SEO professionals approached content like a formula, generating a brief, stuffing a few keywords into templated sections, and hoping the algorithm rewards refreshing the page. That era is over.

Today, with the rise of AI Mode and large language model-powered search, content must demonstrate genuine understanding, not just surface-level optimization. Search engines — and more importantly, users — are looking for substance. Relevance now depends on how well your content aligns with intent, how it contributes to a broader discovery journey, and how meaningfully it answers real human questions.

But this shift requires a fundamental rethinking of how we define a content brief. It can no longer be a checklist of keywords and headers. It must start with a deep understanding of audience expectations, not just query volume. That means going back to the roots of true consumer research, the kind that informs product development, brand strategy, and customer experience design.

When was the last time a content strategist used insights from focus groups, usability testing, support transcripts, or customer panels to inform a brief? Most SEO tools can tell you what’s been searched. Very few can tell you why it’s being searched. AI output requires defining the unmet need beneath the prompt or request, it is the pure answer of why vs what or how. That kind of insight doesn’t come from scraping the SERPs. It comes from listening to customers and users, and being integrated with the real expectation of a business service, product or value proposition. 

We need to bring qualitative insights back into the process. AI enables us to scale research like never before. At Previsible, we recently ran an experiment where we mined hundreds of sales call transcripts and thousands of customer support entries to redesign solution pages – ZERO use of ranking and keyword data was used. This kind of content addressed real-world pain points using the voice of the customer. We added the following to all of our briefs to address the changing landscape of multimodal expectations:

AI can now process audio, video, written feedback, and even social comments at scale. It can extract sentiment, surface patterns, identify recurring frustrations, and cluster related concepts into usable insights. That means SEO research is no longer dependent on other teams and skills to acquire the insights needed to perform optimizations. With access to raw data and inputs, we control the direction of integrating insights directly to customers performing discovery in everything from AI Mode to LLM based prompts.

The modern content brief should include:

• Insights from real customer interactions

• Common objections or pain points from support or sales

• Language patterns from high-performing user reviews

• Behavioral patterns from on-site testing or heatmaps

• Emotional tone and sentiment from qualitative feedback

In this new world, optimization is not just about satisfying simplistic keyword level data, it’s about anticipating interest, intent, and initiative, addressing the fan-out of related queries, and creating content that earns trust and attention.

The next generation of SEO briefs will be grounded in multimodal research and crafted with human-centered depth. This is not the death of content optimization, it is a respawning.

Multimodal: The New Standard for User-Centric Content

This is where multimodality is directing the future of work. In the jobs-to-be-done (JTBD) framework we reframe the connection between a user and user’s problem or needs by phrasing the need as a job that the user wants to get done. Our SEO world is now expanding expectations in production and value by connecting with users in the same way as JTBD meets users needs in product development. 

Google is increasingly integrating beyond the two dimensional conversation to elements that drive enhanced experiences within text, image, video, and real-time visual inputs. These expectations will drive the understanding and purpose of the web page. For years Google has centered the focus on “Content Quality” – this is no longer a valid standard. The standard to bear is how we can add value beyond the quality that persists on a web page. Making elements such as simply “add keywords” or “adding images for SEO” no longer achieve a consumer and comprehensibly informational outcome. 

The development cycles are going to move even faster with the addition of Search Live and Project Astra, where Google will provide users with camera and visual interaction with real time results. This radically expands the context and depth of what search engines need to understand. Making our jobs as SEOs even more important and the need to transcend our frameworks is critical. 

As SEOs, this means your content strategy must consider:

• How your visual assets help explain or demonstrate value
  
• Whether video or audio adds clarity or utility to your topic
  
• How structured data can help search associate and link multimodal content

Multimodal is not just a formatting requirement—it’s a strategic shift toward interactive, contextual, and accessible information experiences.

Experience Quality: A New Foundation For Engagement, Learning, & Action

Multimodality is not a formatting trend, it’s the new foundation for how users engage, learn, and act in digital environments. In SEO, we must reframe our role through the Jobs-to-Be-Done (JTBD) process: users don’t just want information, they want to complete a task, solve a problem, or advance a goal.

Search behavior today reflects this shift. With AI-results and advancements in discovery like Search Live and Project Astra, Google is no longer indexing web pages for content alone; they are evaluating whether your content helps a user complete their need, across a spectrum of modalities: text, images, video, voice, and real-time camera input.

For years, we’ve been optimizing for the somewhat nebulous concept of “Content Quality.” That bar is now table stakes. What matters is experience quality, can your content demonstrate, guide, or assist in a way that matches how users think, ask, and interact?

What Multimodal SEO Requires Today

As SEOs, this means your strategy must now answer:

• Does your content support different learning styles and contexts?
  
  → Use visuals, animations, and diagrams to show—not just tell.
  
• Does audio or video enhance understanding or action?
  
  → Embed walkthroughs, voiceovers, or explainer clips to support complex or high-intent queries.
  
• Is your content machine-readable in all formats?
  
  → Leverage structured data (e.g., schema.org) to clarify relationships between visuals, text, and functionality.
  
• Can your content respond to real-time needs?
  
  → Prepare for live search triggers: product recognition, environment-based queries, or AR overlays.
  
• Is accessibility built-in?
  
  → Include alt text, subtitles, and mobile-first design to support broad and equitable discovery.

Multimodal optimization isn’t a checklist—it’s a strategic upgrade that aligns content with human behavior in a rich, AI-assisted world.

SEO Must Now Serve as the Strategy Behind Research and Experience

Google’s AI Mode, query fan-out, Deep Search, and multimodal capabilities are all converging to create an experience that mimics how humans research and learn. In a recent podcast interview, Google CEO Sundar Pichai outlined the current plan for AI mode which “will remain in a separate tab but over time, Google will keep migrating it to the main page.”Highlighting our everyday access to the discovery platform, Google will integrate AI features and expand the main search experience. Getting ready for this reality is critical to being an “SEO professional” today and in the years to come. 

This means SEO must now support:

• Discoverability across a network of related queries, and ensure that we identify expectations from fan-out research
  
• Content that helps LLMs reason, not just match keywords, making data an engagement aspect of our web pages
  
• Media that complements the way users think, see, and engage
  

If SEO is to stay relevant, it must evolve from a ranking-focused discipline into a content strategy and UX architecture. Query fan-out and multimodal search aren’t features—they’re the new fundamentals.

This is not a change in ranking factors. These are not even ranking factors anymore! It’s a change in how search understands the web, what answers we address, and how we scale the use out of AI in our everyday work. 

SEO Must Now Serve as an Experience Engine

Google’s AI Mode, query fan-out, Deep Search, and multimodal capabilities are converging to create a search experience that mimics how humans research, reason, and learn.

In a recent podcast interview, Google CEO Sundar Pichai confirmed this direction, stating that AI Mode “will remain in a separate tab but over time, we will keep migrating it to the main page.” This isn’t a side experiment—it’s the future of search.

And that future is coming fast.

From Ranking to Resolving

Traditional SEO aimed to win blue links. But AI Mode no longer ranks content—it synthesizes it, pulling pieces from across the web to fulfill complex, layered queries. With query fan-out, a single user prompt can trigger dozens or hundreds of sub-queries, bringing together multiple sources to resolve the user’s full intent.

This means relevance is no longer a binary match—it’s contextual, compositional, and experiential.

From Content Optimization to Experience Design

Your job as an SEO is no longer to make content discoverable. It’s to make content actionable, adaptable, and agent-ready:

• Can your content support AI agents completing tasks?
  
• Is your information structured, real-time, and multimodal?
  
• Will your experience surface within a zero-click journey that still delivers value?
  

If your answer is no, you’re not just behind on SEO—you’re missing the core mechanics of how discovery now works.

From SERP Manipulation to Outcome Enablement

This isn’t the death of SEO. It’s the rebirth of SEO as a full-spectrum experience engine.

Winning tomorrow’s search means building content ecosystems that:

• Serve layered intent and support subtopic depth
  
• Integrate audio, video, images, and structured data to match multimodal expectations
  
• Enable discovery, comprehension, and conversion—without a click
  

The New SEO Mandate

As Google continues integrating AI into the core search experience, getting ready for this reality is no longer optional. It’s the number one job.

SEO must now operate like a product and experience team—feeding structured, useful, and emotionally resonant content into a system that increasingly thinks like a user.

Search is no longer just about what’s indexed. It’s about what helps.

In the years ahead, the most impactful SEO strategies won’t be defined by rankings. They’ll be measured by outcomes delivered, tasks completed, and value experienced—at the moment of need, across any modality, inside an AI-native search experience.

SEO has always adapted. Now, it must transcend.