Hi everyone, this is Neo.

If you’ve been in SEO for more than a week, you know the mantra by heart: search intent comes in three flavors — informational, navigational and transactional. That framework organized our keyword research and content planning for well over a decade.

In 2026, that framework is being dismantled — by Google itself.

Now that AI Mode has crossed one billion monthly users, Google has published its first new intent taxonomy: Explore, Decide, Learn, Create and Do — five verbs, not three nouns. And two fresh studies — an empirical analysis of AI Overviews on Spanish publishers and an eye-tracking experiment — show that search intent really has fractured, and it fractured along predictable lines.

In this post I’m going to stitch both halves together: what Google’s five-intent framework actually means, what the two studies’ data tells you, and how independent site sellers should restructure their content strategy around it.

1. The big picture first: zero-click wasn’t invented by AI — AI just put it on steroids

A lot of people blame AI Overviews for “nobody clicking your site anymore.” But SparkToro’s 2026 clickstream data shows that 68% of Google searches in the U.S. now end without a click to the open web — up from 49% in 2019. The zero-click web is the result of a decade of evolution, not something that appeared out of thin air in 2025.

The real change is that search now happens everywhere: users get answers on YouTube, TikTok and Instagram, ask questions in communities, and order directly inside apps. Google’s answer to people wandering off was to bake the answers directly into the results page.

And AI Mode has taken that to its logical endpoint. Google’s own numbers:

• AI Mode has over 1 billion monthly users, one year after launch, with query volume doubling every quarter;
• The average AI Mode query is 3x longer than a traditional search, packed with action words like “explain” and “summarize,” and full of first-person “I” — users are talking to an assistant, not typing keywords;
• Follow-up questions are growing 40% month over month;
• More than one in six AI Mode searches now includes voice, image or video.

The takeaway is blunt: users no longer want “a link” — they want “a finished answer.”

2. Google’s official answer: search intent goes from “three nouns” to “five verbs”

Facing that reality, Google published a new framework. In its own telling, AI Mode behavior clusters around five actions — each one a completely different kind of need:

Intent	What the user wants	Typical phrasing	Growth signal	What it means for your site
Explore	No fixed goal yet, hunting for inspiration	“ideas for a weekend trip”	Growing 30% faster than AI Mode overall	Evergreen explainers, listicles and inspiration content get pulled into AI answers
Decide	Comparing options, narrowing toward a choice	“which OLED TV should I buy”	“Which”-queries growing 40% faster	Comparison content is the new “category page”
Learn	Understanding a concept, studying, upskilling	“explain quantum computing”	Quiz/study-guide queries surging	Structured explanations and FAQ-style answers earn citations
Create	Generating content, images or assets with AI	“make a logo for my blog”	More than tripled since the start of the year	Assets, templates and tool content are the opening
Do	Booking, planning, completing a task	“book a hotel in Chicago”	Planning queries growing 80% faster	Landing and booking pages are where the clicks actually land

The key insight in this table: these five verbs are not five keyword groups — they’re five tasks. The same topic can hide very different verbs. “How to choose a treadmill” (Decide) and “treadmill training plan” (Do) are different content products even when the keywords look alike.

And here’s the detail most people skip: of the five verbs, only “Do” still visibly clicks out to external sites (booking hotels, buying things, signing up). Most of Explore, Decide and Learn happens inside the AI answer itself — which is why “ranking” is steadily becoming “getting cited.”

3. Study one: Spanish publisher data — which queries actually trigger AI summaries

The first study is an empirical analysis of AI Overviews’ impact on Spanish media (by Clara Soteras, with MJ Cachón and University of Barcelona researchers, under the CUVICOM project). It’s a publisher sample, but several findings are rock-solid for anyone creating content:

1. Longer, more conversational queries trigger AI summaries. Four-word queries trigger an AI Overview 48.1% of the time; three- and four-word queries together account for nearly 70% of all AI-generated results.

2. Question words are almost a guaranteed trigger. The classic “6Ws” of journalism scored strikingly high:

Query starter	AI Overview trigger rate
Why	92.3%
What	85.7%
Who	68.4%
How / Where / When	~60%

In other words, a “Why does X happen” headline is practically a guaranteed AI summary format now.

3. AI loves evergreen, hates breaking news. 75.2% of queries that trigger AI Overviews are evergreen — questions whose answers don’t change for years. During breaking-news moments, the AI Overview trigger rate collapses to 1.1%, barely coexisting with the “Top Stories” module (1.4%).

4. 43.6% of analyzed SERPs show neither an AI Overview nor Top Stories — the authors call this a “blue ocean” where traditional organic ranking still works extremely well. Traditional SEO is not dead; it’s just no longer the whole game.

5. The brand paradox: traffic leader ≠ AI voice leader. On brand-related queries, 20 Minutos (not the biggest publisher in the sample) got 38.9% AI Overview presence, while El Español — the overall traffic leader — sat at 11.1%. Massive visitor volume does not automatically convert into AI search visibility. Who AI cites and who AI trusts run on a completely different logic than who gets the most traffic.

4. Study two: the eye-tracking experiment — there’s a chasm between attention and clicks

The second study is an eye-tracking experiment (Soteras with Diego Criado and MJ Cachón), and it answers “what do users actually look at and click on a SERP?”

The finding is counterintuitive — the researchers call it the “attention-to-action gap”:

• Image modules, product listings and the “Explore Further” button inside AI Overviews capture up to 100% of visual attention while delivering 0% click-through — users scan them in milliseconds and move on;
• The elements that actually convert are the AI Overview’s core answer text (86.4%) and traditional organic links (95.5% CTR) — still the undisputed kings of the SERP.

In plain language: being seen isn’t being clicked, and being clicked isn’t being chosen. AI-era users have developed a form of visual shielding — all the flashy auxiliary modules are noise. Their deliberate attention goes to exactly two places: the AI answer itself, and the organic results near the top.

The lesson for independent sites cuts both ways: the good news is top organic rankings are still click royalty — the fundamentals haven’t expired. The bad news is if you’re neither in the top few organic results nor cited in the AI answer, you’re the noise people scroll past.

5. A content overhaul checklist for independent site sellers

Putting all three pieces together, here’s a concrete new content playbook:

1. Give every piece of content a “verb label.” Before you greenlight anything, ask: does this help someone Explore, Decide, Learn, Create or Do? Don’t let “similar keywords” fool you — the verb dictates the architecture: comparison pages should be built for Decide (“which one should I buy”); tutorials should use Learn’s structured-explanation logic; product pages should quietly serve Do.

2. Push evergreen content toward Q&A structure. If Why-style queries trigger AI summaries 92.3% of the time and evergreen content makes up 75.2% of AI-triggered queries, restructure your evergreen posts as “direct answer first, expansion second”: answer the “why/what/how” head-on in the first 200 words, then unfold the reasoning. It’s citation-friendly formatting for AI and skimmable formatting for humans.

3. Comparison content is your new “decision page.” With Which-queries growing 40% faster than AI Mode overall, users are outsourcing comparisons to AI — and AI compares using your comparison tables, spec pages and genuine reviews. Maintain Decide-content like product pages: accurate specs, fresh data, a clear point of view.

4. Breaking-news and time-sensitive content: don’t chase AI, chase rankings. AI barely shows up in breaking-news moments (1.1%), so that content still wins on classic rankings and speed. And remember the 43.6% blue ocean: on queries where AI is inactive, traditional SEO still has excellent ROI. Don’t let AI-era anxiety make you abandon a channel that still pays.

5. Brand queries are ground zero for “being defined by AI.” The brand paradox shows that big-traffic brands don’t automatically own AI voice. When your brand and product queries trigger an AI summary, whatever the AI says is what users believe — so your brand-information consistency (site, business profile, third-party reviews) is now a content program in its own right. Don’t let the AI define you through someone else’s stale secondhand info.

Neo’s take

This new intent taxonomy announces one thing in essence: the endpoint of search is no longer a link — it’s a decision.

The entire traditional SEO machine — keywords, rankings, clicks — was built around “getting the user to click in.” But across Explore, Decide and Learn, the user may never click anything; the AI has already done half the deciding for them. The content-marketing battlefield is shifting from “position on the results page” to “a seat at the citation table in AI answers.”

That doesn’t mean abandoning the old craft. In fact, quite the opposite:

• The 95.5% CTR on organic links reminds me that the value of ranking in the top spots hasn’t depreciated one bit — don’t cripple yourself out of panic;
• The 43.6% blue-ocean SERPs remind me that AI coverage is nowhere near total — keep winning the queries where it isn’t;
• The 75.2% evergreen trigger rate reminds me that deep, durable content is more valuable in the AI era — because AI only cites what survives verification.

My own read for the rest of 2026: content is no longer a volume game. It’s a game of “when a question is needed by both an AI and a human at the same moment, is your page the one answer?” The five-verb framework is just that sentence translated into an actionable system: each verb is one content production line.

And one sharper angle to close with. Those modules in the eye-tracking study that capture 100% of attention and 0% of clicks — they look a lot like a lot of independent site homepages: flashy, attention-grabbing, forgotten the second the user scrolls past. Attention is transactional; loyalty is the moat. That line comes from the media world (Sean Griffey, co-founder of Industry Dive, calls it “reciprocity”), but it applies to independent sites just as hard: instead of obsessing over squeezing out one more click, ask why a user’s first instinct — when they actually need something — would be to think of you. In the AI era, that “being thought of” status is worth far more than a single click.

Restructure your content architecture around the verbs, then answer one question: if the user never clicks in, can your content still help them decide? Figure that out, and you’ll be fine no matter how AI keeps changing.