Here's a confession that might make me sound ancient: I remember when press releases were actually a legitimate SEO strategy. Not the spray-and-pray wire services that became link farms, but the real craft of building releases around nested keyword phrases. Turns out, that "old school" technique was doing something we now call query fan-out, about two decades before anyone gave it a name.
Greg Jarboe, who's been in the SEO-PR game longer than most of us have had LinkedIn accounts, recently published a piece in Search Engine Journal that made me nearly spit out my morning espresso. He called his 2003 approach "Russian nesting dolls." The idea was simple: find a four-word phrase that contains a three-word phrase inside it. Build your content around the longer version, and you automatically capture searches for both.
His example: "cheap airfare to Philadelphia" contains "airfare to Philadelphia." Use only the shorter phrase, and you're invisible to anyone who types the longer one. Use the nested version, and you catch both.
That's not a clever trick. That's anticipating queries that don't exist yet.
The 15% That Refuses to Shrink
Here's where it gets interesting for those of us obsessing over AI search visibility. Google has maintained for years that roughly 15% of daily queries are ones it has never seen before. John Mueller apparently revisited this number at Search Central Live NYC in March 2025 and seemed almost annoyed by how stubborn it is. As Jarboe recounts, Mueller said he "would have thought at some point most of the searches would have been made; people just ask the same thing over and over again." But the figure just sits there, decade after decade.
Multiply 15% against billions of daily queries, and you're looking at hundreds of millions of searches every single day for phrasing that has literally never existed before. Most of it isn't random gibberish. It's the linguistic offspring of breaking news, new product launches, policy changes, phrases journalists coin on deadline that a thousand people then type into search boxes within the hour.
News creates language. Language creates search volume. And the brands that anticipate that language win.
What AI Search Actually Does Behind the Curtain
Now fast-forward to 2026, and we're watching AI systems do programmatically what Jarboe was doing by hand with press releases. When you ask ChatGPT or Google's AI Mode a question, you're not triggering one search. You're triggering many.
Similarweb's analysis breaks down the mechanics: the AI takes your query, classifies your intent, identifies variables, predicts follow-up questions, generates alternative phrasings, and fires off multiple sub-queries in parallel. A question like "What are the best protein powders for runners?" might spawn searches for "best protein powder for post-run recovery," "whey vs plant-based protein for endurance athletes," "protein supplements for marathon training reviews," and several more.
The final answer synthesizes insights from all of them.
Research from 85SIXTY analyzing over 72,000 AI-generated queries found that a single prompt to ChatGPT or Gemini routinely triggers 8 to 10 parallel, hyper-specific queries before returning an answer. Here's the kicker: 95% of those fan-out phrases show zero monthly search volume in traditional keyword tools. They're invisible to conventional SEO, yet they're the gatekeepers of generative visibility.
The Length Paradox Nobody's Talking About
Traditional Google searches average around 3.4 words. Fan-out queries from ChatGPT average 5.5 words. Gemini's average 9.1 words. As 85SIXTY's analysis puts it, long-tail is no longer optional. It's the primary battlefield.
This is where Jarboe's 2003 framework becomes prophetic rather than nostalgic. He was building content for queries that didn't exist yet by nesting phrases inside phrases. AI systems are now generating those queries automatically, and they're longer, more specific, and more numerous than anything we optimized for in the keyword-density era.
Conductor's breakdown of the phenomenon puts it bluntly:
"If your content doesn't address these hidden layers of intent, you risk being ignored by the very models your customers are using."
What This Means for B2B Marketing Strategy
Let me translate this into boardroom language, because I know half of you are already thinking about how to explain this to your CEO.
First, your keyword strategy is probably too shallow. If you're still optimizing for head terms and hoping to catch long-tail traffic as a bonus, you're playing a game that ended around 2024. AI systems are actively generating the long-tail queries themselves, and they're looking for content that answers those specific formulations.
Second, timeliness matters more than ever. Jarboe's insight about press releases wasn't just about nested phrases. It was about being first to publish content around emerging language. When a new product category emerges, when regulations change, when your industry gets disrupted, the brands that publish thoughtful content within hours (not weeks) are the ones that get indexed and cited when AI systems start fielding questions about those topics.
Third, topic clusters aren't just an SEO tactic anymore. They're survival infrastructure. Conductor recommends building content ecosystems where pillar pages link to detailed subtopic pages, each addressing specific angles that fan-out queries might target. The AI isn't looking for one perfect page. It's building answers from a constellation of sources.
The Uncomfortable Truth About Measurement
Here's where I have to be honest with you: measuring this is hard. Traditional rank tracking tells you where you stand for queries people actually type. It tells you nothing about whether you're showing up in the sub-queries AI systems generate behind the scenes.
Industry data suggests that high-consideration B2B categories see deeper interrogation from AI systems, sometimes 10+ sub-queries per prompt. Lower-consideration categories still face 6 or more. The AI is doing due diligence, cross-checking sources, looking for consensus, time-stamping knowledge, and risk-balancing options before it commits to an answer.
If your brand doesn't survive that cross-examination, it doesn't make the final cut. And you might never know it happened.
Playing the Long Game
Twenty-three years ago, a PR guy in the SEO trenches figured out that building content around nested phrases would capture searches that hadn't been typed yet. He called it Russian nesting dolls. We now call it anticipating query fan-out.
The tools have changed. The scale has changed. The stakes have definitely changed. But the core insight remains: the brands that win in search are the ones building for the questions that don't exist yet, not just the ones being asked today.
Data tells you the what. But anticipation tells you the what's next. And in a world where AI systems are generating hundreds of millions of novel queries daily, that's the only game worth playing.
