Google AI Mode Data Proves the Keyword Era Is Over

Google just published a year of AI Mode usage data. The average query is now triple the length of a traditional search. The third most common opening word is "I." One in six searches uses voice or images instead of typed text. The persona your content strategy was built around last summer no longer describes what over a billion people are actually doing when they search.

The Keyword User Is Now the Minority

Here is what most content teams still assume: a person types three or four words into Google, scans the results, clicks something. That assumption guided every keyword strategy shipped in 2025. A Nielsen Norman Group study confirmed the behavioral shift is real but noted that old habits are sticky, which means most teams are still running their 2025 playbook while their users have already moved on.

Google's own data says that description now covers a shrinking share of how AI Mode users actually behave. Shivani Mohan, VP of Data Science and UXR at Google Search, laid out the numbers in a report covering AI Mode's first year in the U.S. (May 2025 through April 2026). Queries have more than doubled every quarter since launch. Follow-up queries have grown more than 40% per month on average, meaning users are not landing on one answer and leaving. They are staying in the conversation and going deeper.

The top five opening words in AI Mode searches are "what," "how," "I," "is," and "can." Look at that third word. "I." People are narrating personal context into the search bar. Not "running shoes flat feet." Something closer to "I have flat feet and my knees hurt after every run, can you find me a shoe that will not make it worse?"

Google's own example is blunter: "I hate cardio. Give me a routine that avoids it but still works."

That is not a keyword. That is a person talking to someone who might actually help them.

Three Behavioral Shifts the Data Confirms

1. Queries became conversations. The average AI Mode query is three times the length of a traditional search query. The top five keyword themes are Information, Identify, Find, Explain, Summarize. Every one of those is a verb requesting synthesis, not a noun requesting a page. Your H2 structure optimized for "[best X 2025]" does not answer what the AI Mode user is actually asking.

2. Search is no longer text-only. More than one in six AI Mode searches in the U.S. now uses voice or images. Image-input searches are growing more than 40% month over month. Google Lens alone now has over 1.5 billion monthly users searching with their camera. Someone photographs a product and asks AI Mode what it is, where to buy it, whether the cheaper version works. Your alt text written for accessibility compliance and your alt text written for a multimodal search user are two different things.

3. AI Mode became a decision tool, not a discovery layer. Planning queries in AI Mode have grown 80% faster than the overall query pace over the past six months. Brainstorming queries grew 30% faster. Searches starting with "which" have grown 40% faster since launch. People are not using AI Mode to browse. They are using it to choose. If your content does not help the machine help the user decide, someone else's content does.

What to Do About It

Rewrite your top pages for the conversational version of the question. Take the primary keyword for each of your ten highest-traffic pages. Rewrite it as a full sentence the way an AI Mode user would actually say it. If your content does not answer the long-form version, a competitor who does answer it will eventually replace you in the response.

Build a follow-up question inventory. The 40%-per-month growth in follow-up queries means the second and third question matter as much as the first. If you know what your most common entry-point queries are, the follow-up question is now strategically as important as the entry. Most teams do not have this inventory.

Prepare your visual assets for multimodal queries. One in six queries is already non-text. Image search is the fastest-growing query type in the system. Product images, diagrams, and screenshots need structured context that serves a user who photographed something and wants an answer, not just a screen reader.

Where Machine Relations Enters

When a person types "best CRM 2025," the machine returns a list. When a person says "I run a 12-person agency and our current CRM can not track which prospects came from earned media versus paid, what should I switch to?" the machine needs to trust a source before it recommends.

That is the shift. The keyword era rewarded relevance. The conversational era rewards authority. The machine will not put your product in someone's answer unless it has evidence you are worth citing. Earned media, structured data, entity consistency, and source architecture are the materials that build that trust. That is Machine Relations: the discipline of becoming a source the machine will cite when a real person asks a real question.

Google has over 1 billion monthly active users on AI Mode. Queries are doubling every quarter. The behavioral shift is not a forecast. It is a data set with a year of receipts.

Your keyword strategy is optimizing for a user who is disappearing. The user who replaced them talks in full sentences, uploads photos, asks follow-up questions, and expects the machine to help them decide. Your content either answers that person or it does not.

There is no middle position.

FAQ

How many people use Google AI Mode?

Google reported over 1 billion monthly active users globally as of May 2026. Queries have more than doubled every quarter since the U.S. launch in May 2025.

Are keywords still relevant for AI Mode optimization?

Keywords still matter as topic signals, but the search behavior has changed. The average AI Mode query is three times longer than a traditional search. Optimizing for three-word keyword strings alone misses the conversational, multimodal queries that are growing fastest.

What is multimodal search in AI Mode?

Multimodal search means using voice, images, or video as input instead of typed text. One in six AI Mode searches in the U.S. already uses a non-text input, and image searches are growing over 40% month over month.