For two decades, before the industry coined the term "query fan-out," many seasoned SEO professionals were practicing a silent, intuitive craft. They called it something else—the "Russian nesting doll" strategy. It was a method born of necessity, long before generative AI or search engines became the complex, multi-modal ecosystems they are today.
Today, as search behavior shifts from simple keyword matching to conversational, AI-driven exploration, that old-school instinct has become a competitive necessity. Understanding how to capture the "unseen" 15% of search queries is no longer just about keyword density; it is about architectural precision in how we construct content.
The Core Concept: The Anatomy of the Russian Nesting Doll
The "nesting doll" approach to search optimization is elegant in its simplicity. When you optimize for a three-word phrase, you are effectively closing the door on any searcher who uses a four-word variant. However, by optimizing for the longer, more specific four-word phrase that naturally contains the three-word core, you occupy both spaces.
Consider a simple example: "airfare to Philadelphia." If your content only mentions this three-word string, you are invisible to the user who types "cheap airfare to Philadelphia." By constructing your sentences to include the modifier—the "outer shell" of the doll—you capture the broad query and the granular one simultaneously.
This is not a "trick." It is a structural alignment with the reality of how search engines handle linguistic variability. Most SEOs undervalue this because they focus on existing search volume. The real opportunity lies in the queries that nobody has typed yet—the 15% of search volume that remains elusive.
Chronology: From BERT to the AI Search Era
The significance of the "unseen" query was first formalized by Google in 2019 with the introduction of BERT (Bidirectional Encoder Representations from Transformers). Google revealed that 15% of the queries it processes daily have never been seen before.
For years, many analysts assumed that as search engines matured, this figure would decline. The logic followed that as more people searched, the "long tail" would eventually be fully mapped. Yet, in March 2025, at Search Central Live NYC, Google’s John Mueller addressed the persistence of this statistic with a note of mild frustration.
"I would have thought at some point most of the searches would have been made; people just ask the same thing over and over again," Mueller noted. "But when we recalculate these metrics, it’s always around 15%."
Even with the integration of Large Language Models (LLMs), that 15% remains a constant, immovable fixture. This reveals a critical truth: news, social trends, evolving product names, and journalistic shorthand create new language daily. Search volume is a living, breathing entity.
Supporting Data: The Mechanics of Fan-Out
In August 2024, MJ Cachón published a groundbreaking dataset study that quantified the phenomenon of "query fan-out." By running 189 branded prompts through ChatGPT, Cachón observed the model firing off 1,797 sub-queries that were never explicitly typed by a human user.
The AI Narrowing Process
Cachón’s data revealed that AI systems do not behave randomly when they fan out. The process is a reverse-nesting progression:
- The Seed Query: The process begins with conversational, broad language.
- The Operator Phase: The AI narrows the scope using specific commands like
site:operators. - The Verification Phase: The system pulls exact quoted phrases to verify that a source matches the intent.
Cachón’s research showed that quote usage within these AI-generated sub-queries climbed 25-fold from the first search to the last. This validates the "nesting" theory: AI systems are drilling down from broad concepts to literal, verifiable text. If your content lacks the exact wording at multiple levels of specificity, the AI system will bypass your page entirely, favoring sources that provide a "verifiable quote."
Furthermore, as of May 2026, internal data suggests that AI-mode queries are now, on average, triple the length of traditional search queries. We are moving toward a paradigm where length is not a side effect of AI search; it is the terrain itself.
Implications for Modern SEO and Content Strategy
The industry spent the 2010s obsessed with "head terms"—high-volume, high-competition keywords that dominated budget meetings. This was a backward approach even before the AI revolution, and it is catastrophic now.
When a single user prompt fans out into a dozen specific, seven-word sub-queries, your content must be versatile enough to serve all of them. If you are only ranking for the seed term, you are missing out on the majority of the "fan-out" traffic that occurs in the background of the user’s experience.
The Shift to "News-Speed" Content
Press releases and rapid-response content remain the most effective tools for capturing the 15% of new queries. Because these formats are published at the same speed as the news cycle, they can "own" a phrase the moment it enters the lexicon. A blog post often arrives too late to capture the initial surge of a new term. To be effective, an organization must transition from a static content calendar to a "speed-to-language" model.
Applying the Strategy: Three Habits for the AI-First World
You do not need sophisticated API access to leverage these insights. Success in the current search environment requires adopting three specific habits:
1. The Nesting Habit
Never stop at the seed phrase. When conducting keyword research, identify the core three-word term, then build a "nest" around it by identifying the four- and five-word phrases that naturally contain it. Structure your H2 headers and opening paragraphs around these longer, more descriptive variants. Use tools like Google Search Console to monitor high-impression, low-click queries; these are the long-tail variants that are already knocking on your door, waiting for you to optimize for them.
2. Speed-to-Language
Publishing must be decoupled from long-lead editorial calendars. The most valuable search queries—the ones that define new trends—are born from breaking news. If your brand can produce a piece of content that uses the exact, nascent terminology of a breaking event, you capture the search volume before competitors even realize the search demand exists.
3. The "Quotable Answer" Principle
Cachón’s data confirms that AI systems verify claims by searching for exact quoted phrases. This means your content must be written with "atomic" clarity. If an AI cannot lift a single sentence from your article and use it as a complete, coherent answer to a query, you have failed the optimization test. Rewrite your core answers until they are standalone, declarative, and easily quotable.
Conclusion: Returning to the Fundamentals
The shift toward generative search and AI-driven query fan-out might feel like a radical departure, but it is actually a return to the roots of information retrieval. The "Russian nesting doll" strategy proves that the best way to remain visible is to be comprehensive.
By building content that serves both the broad intent and the granular, "unseen" long-tail queries, you future-proof your digital presence. The AI era doesn’t require us to invent new tricks; it requires us to be more deliberate about the ones we have ignored for too long. The 15% of unseen queries are not a mystery—they are the next wave of traffic, waiting for the content that is structured specifically to catch them.
