Google's AI Mode Has 1 Billion Users — And Their Queries Are 3x Longer Than Yours

Most SEO teams spent 2025 optimizing for three- and four-word queries. That's what the keyword tools showed. That's what the guides said. That was fine — until Google released actual behavioral data that makes a lot of that work look like it was targeting the wrong person entirely.

In May 2026, Google's VP of Data Science & UXR for Search published a report called "How People Are Using AI Mode in the U.S." — covering the period from AI Mode's US launch in May 2025 through April 2026. It's not a blog post of vague platitudes. It has specific numbers, and those numbers are a problem for content teams that haven't adjusted yet.

The headline stat: the average AI Mode query is now triple the length of a traditional Google search query. That's not a marginal shift. That single figure means the person your content was built for — the one typing "best running shoes 2025" and scanning a listicle — is a shrinking fraction of how people actually arrive at answers in AI Mode.

What Google's Own Data Actually Says

The report covers a year of AI Mode behavior, and the numbers stack up in one direction. Here's what's happening:

• Queries are 3x longer on average than traditional search.
• Follow-up queries grew 40%+ per month — users stay in the conversation and go deeper, not land-and-leave.
• 1 in 6 AI Mode searches is now multimodal — that means voice, image, or video input rather than typed text.
• Image-input searches grew 40%+ month over month since launch.
• Planning queries grew 80% faster than the overall AI Mode query pace.
• Queries beginning with "which" — decision-making queries — grew 40% faster over the last six months.

And here's the one that should make content strategists pause: the top five opening words in AI Mode searches are "what," "how," "I," "is," and "can." That third entry — "I" — tells you something important. People are narrating personal context into the search bar.

Not "Running Shoes for Flat Feet" Anymore

The example Google used in the report is blunt and worth quoting directly: instead of searching "running shoes flat feet," an AI Mode user types something closer to "I have flat feet and my knees hurt, can you help me find a running shoe that won't make it worse?" In the health category, it gets even more conversational: "I hate cardio. Give me a routine that avoids it but still works."

That's not a keyword. That's a person talking to someone who might actually help them. Most content built for the former can't serve the latter, even if the underlying topic is identical.

The Content Gap Is Real — and Most Teams Haven't Seen It Yet

Google organizes AI Mode behavior into five categories: Explore, Decide, Learn, Create, and Do. The fastest-growing ones are the deliberate ones. Brainstorming queries grew 30% faster than overall AI Mode query volume. Planning queries grew 80% faster. "Which" queries — the ones where someone is genuinely trying to make a decision — grew 40% faster over six months.

This is where the content gap actually lives. If you wrote a "best running shoes" roundup optimized for a short-tail keyword, that page serves someone scanning for a quick list. It doesn't serve someone who's training for their first 5K, has never bought running shoes before, and wants to know which pair to start with and how they'd know if they fit right.

Both queries express shoe-buying intent. But only one of them describes what AI Mode users are actually doing. And right now, most content — even good content — only answers the first one.

Why "I" Is the Most Important Word in Search Right Now

The fact that "I" is the third most common opening word in AI Mode queries isn't a quirk. It means users are front-loading personal context before asking their question. That's a fundamentally different behavior from traditional search, where you strip out everything personal and boil the query down to its most searchable form.

For content teams, this has a specific implication: your pages need to acknowledge the variety of situations your reader might be coming from, not just answer the sterile version of the question. A guide on setting up Google Search Console for the first time reads very differently depending on whether the reader is a solo blogger, an e-commerce site owner, or an agency taking on a new client. AI Mode users are telling Google which version they are, right in the query.

This doesn't mean you need to write every page for every persona. It means the more context you pack into your content — the more you address who needs this information and under what circumstances — the better you serve the user who's already told AI Mode exactly who they are.

Follow-Up Queries Are Now a First-Class Signal

Here's the data point that most content analyses have glossed over: AI Mode follow-up queries grew 40%+ per month throughout the year. That's not small. It means users aren't getting a complete answer and leaving — they're getting an answer, finding it useful, and then immediately asking something adjacent.

For traditional SEO, your job was to rank for the entry-point query and get the click. Done. But in AI Mode, the conversation continues and your content either participates in that continuation or it doesn't.

What to Do With This

Think about your most common entry-point questions — the queries that bring people to your top pages. Now ask: what would someone reasonably ask next? What's the follow-up to the answer you just gave them?

Most sites don't have a systematic answer to that. They have individual pages. AI Mode has a conversation. The gap between those two things is where a lot of traffic will quietly move over the next 12 months.

Some practical moves:

• Add FAQ sections to your highest-traffic pages that address obvious follow-ups.
• Link internally to the next logical question — not just related topics, but the actual next step in the user's thought process.
• Check your Search Console data for queries where your page ranks but the click-through rate is low. That often signals a mismatch between what users are asking and what your meta description promises.

Multimodal Search Isn't "Coming" — It's Already Here

1 in 6 AI Mode searches is now non-text. Image-input searches are the fastest-growing query type in the system, up 40%+ month over month. That's not a future consideration. That's a current one.

What this means practically: alt text written for accessibility compliance is different from alt text written for a user who's photographed a product and asked AI Mode what it is and where to buy it. The context around your product images, how-to diagrams, and screenshots needs to serve both use cases.

A few things worth checking right now:

• Are your product images described in enough detail in alt text and surrounding copy that AI could connect them to a "what is this?" query?
• Do your how-to articles have images with descriptive filenames and alt text that explain the step, not just label the screenshot?
• Is your structured data (especially for products, recipes, and how-tos) complete enough to give AI Mode the context it needs to cite your content visually?

These aren't advanced tactics. They're table stakes that a lot of teams haven't touched since they set their original alt text rules years ago.

The Scale of What You're Dealing With

One more number to calibrate against: Google AI Mode now has over 1 billion monthly active users globally, and query volume has doubled every quarter since launch. This isn't a feature being tested on a small percentage of searches. It's the primary search interface for a significant and rapidly growing share of users.

The behavioral shift Google documented isn't a forecast. It's a dataset. These users exist now. They're searching now. And most of the content targeting them was written for a different version of that person — the one who typed three words and scanned a list.

That version of the user is still out there. But they're no longer the only person you're writing for, and in AI Mode, they might not even be the majority.

Bottom Line

You don't need to throw out your content strategy. You need to audit it against a different question: does this page answer what an AI Mode user would actually type, or does it only answer the keyword-stripped version of that question?

Start with your top 10 traffic pages. For each one, write down the real-language version of the primary query — the way someone would ask it in full sentences with personal context. Then read your page through that lens. Most of the time, you'll find the gap immediately: the page answers the surface question but doesn't address the situation behind it.

That gap is your content work for the next six months. The data Google published tells you exactly where the users are going. The only question is whether your content is waiting for them when they get there.