Organizing content for AI search: A 3-level framework

Traditional search engines rely on countless ranking signals to deliver the most relevant answers to users.

In the age of AI, signals appear to be simpler – at least for now. 

Court documents from Google’s antitrust case reveal that AI Overviews rely on lighter signals and fewer documents in the index.

However, content clarity and topical connections are more important than ever.  

To be visible in the era of AI search requires content that is: 

• Well-structured, so that relationships between topics are clear.
• Expertly written, so that the meaning and depth are obvious.
• Machine-readable, so entities and relationships are easy to interpret.

In other words, the same framework that has guided SEO for years – architecture, expert content, and optimization – is just as important today. 

This article shows how each can reinforce the signals that matter most in AI search. 

FastSearch and RankEmbed: A quick overview

In U.S. v. Google LLC, the court describes how Google’s FastSearch is used in grounding Google’s generative AI answers. 

In short, the technology retrieves only a subset of search results and relies on lighter ranking signals.

From the court documents: 

• “To ground its Gemini models, Google uses a proprietary technology called FastSearch. … FastSearch is based on RankEmbed signals – a set of search ranking signals – and generates abbreviated, ranked web results that a model can use to produce a grounded response. …FastSearch delivers results more quickly than Search because it retrieves fewer documents, but the resulting quality is lower than Search’s fully ranked web results.”

This could explain why some AI Overviews have surfaced low-quality content – a point that Ryan Jones made on LinkedIn:

The court describes RankEmbed as one of Google’s “top-level” deep-learning signals, capable of “finding and exploiting patterns in vast data sets.” 

RankEmbed is designed to capture meaning by recognizing semantic relationships between queries and documents. 

Dig deeper: The ABCs of Google ranking signals: What top search engineers revealed

Unlike other signals that measure popularity or count backlinks, RankEmbed focuses on how closely a piece of content aligns with the meaning of what the user asked. 

And RankEmbed is a key part of FastSearch.

This could explain why FastSearch, which grounds AI Overviews, sometimes surfaces results that look different from traditional search rankings.

It prioritizes semantic clarity over other authority-type signals.

If RankEmbed is central to FastSearch, then the things that matter to AI visibility are those that strengthen semantic connections.

And those are the strategies I’m going to talk about next. 

1. The foundation: A solid content architecture

How you organize the content on a site creates relationships, and those relationships make it easier for both search engines and AI systems to interpret relevance.  

And at the same time, a clear information architecture allows website visitors to engage more deeply with a website, as it provides complete answers to questions along their search journey. 

This is not a new concept. SEO siloing is something we’ve been doing at my agency for 25 years. 

However, in the age of AI, it helps signal that a site is semantically clear and contextually relevant.

So what’s it all about? Let’s look quickly at this strategy in action. 

Analyzing the business

The first step seems simple, but it’s surprising how much clarity it can offer when organizing a website’s information. 

This step consists of analyzing the products, services, and types of information a brand provides. 

We take this information as a starting point for more research. 

That research includes looking at what topics the site is currently gaining visibility for in search results and what queries are triggering that. 

Then, we look at missed opportunities. If the website were a book, what story would it tell? What chapters would naturally fit under that story? And what chapters are missing?

At the end of this research, you have:

• A roadmap for the structure.
• An understanding of how the content that’s already published fits in. 
• A plan for the topics that need to be created. 

Implementing the structure

A “silo” is the name for the structure. We call it a silo because it keeps topical themes tightly connected without being intermingled with other topics that dilute its relevance. 

When topics are intermingled, it is confusing to users and engines. But when you begin to organize them, it creates clarity. 

The structure can be formed in two distinct ways: either through the physical directory (URL structure) or virtual connections (internal linking).

Let’s look closer at those two tactics now. 

Physical silos 

Physical silos structure URLs that create a hierarchy of topical relevance. 

If a company’s main focus is CRM software, the overarching theme would be “CRM Technology.” 

Under that, categories could be based on the major functions or use cases customers care about, such as:

• Sales automation.
• Customer support tools.
• Analytics and reporting.

That’s three silos. Then, each category gets its own landing page, and is supported by subpages/subcategories that cover specific features or solutions. 

For example, the sales automation silo might include:

• Main landing page: companycrm.com/crm/sales-automation
• Subcategories:
  • companycrm.com/crm/sales-automation/lead-scoring
  • companycrm.com/crm/sales-automation/email-tracking
  • companycrm.com/crm/sales-automation/opportunity-management

For this fictional site, this structure would signal to search engines/LLMs that the site is a good source on CRM software. 

Virtual silos 

Virtual silos, on the other hand, connect related pages through internal links, even if they’re not in the same directory.

This is a good fit when the physical directory cannot be altered in any way.

It’s also an effective hybrid approach that forms relationships between content when the content lives outside of the physical directory, such as in a blog.

For instance, on a CRM website, a blog post about improving customer retention could link directly to the main customer support tools landing page, even though the blog lives in a separate /blog/ directory.

These contextual links build virtual silos.  

Even if the blog isn’t physically nested under /crm/, the internal linking ties it back to the main theme and strengthens topical authority.

It also allows website visitors more opportunities to follow links and engage more deeply with the site as they continue on their customer journey. 

The significance of siloing on LLMs

AI models are more likely to retrieve content that shows clear topic coverage and surface sites that demonstrate semantic consistency across related pages.

However, a bunch of content on a topic spread across the site is not going to build that authority on its own. It has to match intent and be linked properly. 

The siloing strategy may seem quite simple on the surface. However, there are still many technical considerations. 

Then there’s the matter of disruption. Siloing a large website is no easy feat. This is where partnering with an expert can be critical. 

2. Authority layer: True expertise

We are living in a world where AI content is starting to dominate the search results. 

Some speculate that it won’t be long before expertly written, human-generated content will come at a premium. 

And if that’s the case, it won’t be hard to stand out among the generic, machine-generated content if you put in the effort.    

When creating content for your silos, each page needs to answer a query expertly, completely, and with additional resources to other areas of the site. 

Think about how you can elevate your content, not just make it more efficient. In the age of AI, this will once again become a real challenge for many. 

But we still have guideposts for doing this. So, let’s look at the baseline content strategies that will help a brand remain an expert.

Approach writing professionally

Hire people who are professional writers, even if they’re working with you to refine AI content. 

They should inherently have a grasp of how to write well, and will navigate things like:

• Logical consistencies: Make sure to resolve any contradictions or conflicts in your content by thoroughly researching the topic and reviewing contradictory statements.
• Persuasive writing: You can always strengthen your arguments and ensure they are well-supported by using solid research and relevant examples.
• Accuracy: Be sure to verify the accuracy of your information through multiple reliable sources before publishing. Fact-checking is essential to avoid spreading misinformation.
• Ethical standards: Familiarize yourself with and adhere to ethical standards, including not plagiarizing and following search engine guidelines. Ethics around AI is a developing topic that SEO/GEO professionals should familiarize themselves with. 

Dig deeper: Mastering content quality: The ultimate guide

Create helpful content, per Google

By now, most are familiar with Google’s guidance on helpful content, which includes useful self-assessment questions, such as:

• Does the content provide original information, reporting, research, or analysis?
• Does the content provide a substantial, complete, or comprehensive description of the topic?
• Does the content provide insightful analysis or interesting information that is beyond the obvious?
• If the content draws on other sources, does it avoid simply copying or rewriting those sources and instead provide substantial additional value and originality?
• Does the content provide substantial value when compared to other pages in search results?

Marketing leaders should treat these as foundational guidelines during any editorial review of content produced by an SEO/GEO program.

Uphold Google’s E-E-A-T framework

E-E-A-T is a holistic, quality framework for content.

But E-E-A-T must be earned over time through consistently delivering value and building genuine trust with your audience. 

And it matters most for “Your Money or Your Life” (YMYL) topics.

You can earn E-E-A-T by things like:

• Offering original insights or analysis: Your content should provide unique, valuable perspectives outside of what everyone else is saying.
• Demonstrating firsthand experience or expertise: Authentic credentials and real-world knowledge matter. Incorporate anecdotes and professional guidance freely.
• Aligning with user intent: Focus on solving real people’s problems, not just chasing keywords and mass-producing content.
• Avoiding superficial templates or generic output: Authentic, thoughtfully crafted content always wins over a formulaic output (like content often coming from machines).

The bottom line is simple. Write for people first, and build trust over time. 

Ultimately, it all comes down to one thing: creating good results for searchers.

Significance of quality content on LLMs

We know that being a subject matter expert is vital to ranking high in search. And we know that good SEO is needed for good GEO – at least that’s what Google’s Danny Sullivan says. 

At his keynote at WordCamp U.S. in August 2025, Sullivan reiterated that:

• “SEO means you understand how people search for content and then you understand how to have your content there.”

He added:

• “Good SEO is really having good content for people.” 

Remember that while not a direct correlation all the time, studies show there’s often overlap between high-ranking search results and AI Overview inclusions, even though the underlying systems use different signals. 

In other words, there’s a good chance that if you are trusted in search, you will be trusted for inclusion in Google’s generative AI answers.

Some may argue that bad content written by machines is already visible in AI Overviews, so what’s the use of putting in the effort? 

Well, first, because of ethics. We should uphold ethics in our marketing. 

This has been my personal belief for decades, dating back to when black hats were gaming the system every day and winning. 

Secondly, things can turn on a dime. Algorithms can change and content can be wiped from the face of the search results. Let’s not forget the March 2024 updates. 

And in that moment, if you’ve done everything right, you have just been catapulted to the top, a spot that will not easily be taken back. 

3. Signal layer: Structured data/schema

Structured data or schema is the layer that can help translate your content into signals that machines can better interpret. 

This can make it easier for AI systems to connect the dots.

That said, some research suggests that it may not play a role in direct AI Overview visibility.

Even so, Google advises using structured data to ensure content performs well in Google’s AI experiences, and so it’s one of those practices that require experimentation.

Here’s what Google says: 

Many in the industry are already implementing or planning to implement schema/structured data as part of their GEO strategies.  

The SEOFOMO State of AI Search Optimization Survey, 2025 Edition, shows that structured data/schema was most frequently mentioned as a way to optimize for AI search. 

The challenge will be to implement the schema methodically. 

For instance, don’t just use schema on your homepage or a few products; add it everywhere it makes sense. 

However, don’t misuse, abuse, or overdo it. 

Structured data should accurately represent the page’s main content, so choosing the most relevant type of structured data for the content is key.

(Remember that Google states structured data issues can trigger a manual action.)  

And above all, test and validate. 

Significance of structured data on LLMs

AI Overviews rely on understanding entities like people, places, products, organizations and concepts. 

Structured data helps define those entities and makes it easier for search engines like Google (and its AI-powered technology) to trust the information.

Schema.org has been around for almost 15 years, so while it’s not a new tactic, it’s useful for clarifying content, even if its impact on AI visibility is still being tested.

Either way, structured data is resurging as a way to reduce ambiguity in the era of AI search.  

Build clarity for lasting visibility

What does it really take to stand out in AI-powered search? The answer is clarity. 

Clear structure, expertise, and signals help both your audience and search technologies connect the dots.

This is the kind of groundwork that search engines and AI systems depend on. 

The bottom line is that you don’t need to chase every new GEO trick to succeed. 

The fundamentals that have guided SEO for decades are still the path forward. 

Focusing on a site’s information architecture, creating expert content, and using key optimization techniques like schema helps create connections that people, search engines, and AI systems can rely on. 

Dig deeper: Chunk, cite, clarify, build: A content framework for AI search