In its first-year AI Mode usage report, Google says the average AI Mode search runs about triple the length of a classic query, more than one in six U.S. searches now use voice or images, and follow-up questions are climbing more than 40% a month.
2026-10-30 · by Moe Ameen
On May 19, 2026, Google published "How AI Mode is changing and expanding the way people search," a usage report written by Shivani Mohan, VP of Data Science and UXR, marking roughly a year of AI Mode. The headline framing: people are no longer compressing questions into a three-to-five-word keyword string. "People are no longer worrying about the 'right way' to formulate their questions," the report says, and the data it shares is about how the shape of a typical search has changed.
The numbers Google disclosed are directional but consistent. The average AI Mode search is about triple the length of a traditional Search query. More than one in six searches in the U.S. now use voice or images rather than typed text, with image searches growing more than 40% month over month. Sessions have also become more conversational: follow-up queries in AI Mode have grown by more than 40% on average per month in the U.S. since launch. Google says AI Mode has passed a billion monthly active users globally and that its query volume has more than doubled every quarter since launch.
The report also details how the wording of queries is shifting. An accompanying infographic lists the most common opening words in AI Mode as "What," "How," "I," "Is," and "Can," and the most common embedded terms as "information," "identify," "find," "explain," and "summarize" — the vocabulary of full questions, not keyword fragments. Drawing on sampled Google Trends data, Google reports that planning-type queries grew about 80% faster than AI Mode queries overall over six months, brainstorming queries about 30% faster since launch, and decision-stage "which" queries about 40% faster than the average — with "which of" and "which one" among the fastest-growing phrasings. Searches beginning with "where to," "where should I," and "ideas for" are rising too.
Two caveats worth keeping. The figures come from Google's own report, which promotes its products, and the keyword breakdowns are drawn from sampled Trends data, so treat them as the direction of travel rather than precise measurements. And this is a U.S.-focused snapshot from May 2026 — it remains the most detailed query-behavior data Google has published, but the exact proportions keep moving.
This report is really a brief for the kind of content that now wins: answer-first pieces that resolve one specific, question-shaped search — not keyword pages built for a two-word phrase. The practical problem is volume. If the queries worth capturing are long, specific, and span planning, "which," and how-to intent, you cannot cover that surface with a handful of pages; you need a steady stream of genuinely useful content, each one answering a real question, published where both people and AI crawlers can find it.
That is the gap [Kompozy](/) fills. Point it at one source — a video, a customer question, a topic — and it generates a full answer-shaped set under one [Persona Brief](/glossary/persona-brief): a Blog Article structured as a direct answer to a specific question (exactly the format AI Mode's longer, question-shaped queries pull from), an Email Newsletter, plus the native social set — persona video, brand-exact [Carousel Posts](/glossary/hyperframes), Photo Posts, and Quote Graphics — so the same answer shows up across eight social platforms plus your blog and email. Run it on [Autopilot](/glossary/autopilot) behind a per-post review gate and you produce the breadth this shift rewards without hand-building every page. For the strategy side, see our glossary entries on [answer-engine optimization](/glossary/answer-engine-optimization) and [generative-engine optimization](/glossary/generative-engine-optimization), and our reporting on [Google's query fan-out technique](/news/google-r4t-diffusion-query-fan-out) that makes the underlying keyword invisible.
According to Google's May 2026 AI Mode usage report, the average AI Mode search is about triple the length of a traditional query, and people phrase searches as full questions rather than two-to-five-word keyword strings. More than one in six U.S. searches now use voice or images, and follow-up questions within a session have grown more than 40% per month on average since launch. The most common opening words are "What," "How," "I," "Is," and "Can."
Drawing on sampled Google Trends data, Google reports that planning-type queries grew about 80% faster than AI Mode queries overall over six months, brainstorming queries about 30% faster since launch, and decision-stage "which" queries about 40% faster than the average — with "which of" and "which one" among the fastest-growing phrasings. Searches starting with "where to," "where should I," and "ideas for" are also rising. These are directional figures from Google's own report.
Build content around specific, question-shaped searches instead of short keyword phrases. Longer queries carry more intent, so answer-first pages that fully resolve one detailed question — comparisons for "which" queries, step-by-step guides for how-to intent, planning content for "ideas for" searches — are what AI Search pulls from. Because AI Mode fans each question out into sub-queries you never see, the winning strategy is breadth and genuine usefulness across many pieces, not chasing a single keyword.
Google published it on May 19, 2026 in a blog post and downloadable report, "How AI Mode is changing and expanding the way people search," authored by VP of Data Science and UXR Shivani Mohan. The figures are Google's own and the keyword breakdowns come from sampled Trends data, so treat them as the direction of travel rather than exact measurements. It is a U.S.-focused snapshot but remains the most detailed query-behavior data Google has published.