Google Data Shows AI Search Users Moved Past Keywords, Your Content Hasn’t

Google's Own Data Says Search Queries Got Longer and Chattier — It Says Nothing About Who's Writing the Answers

On May 19, 2026, Google published "How People Are Using AI Mode in the U.S." on its Keyword blog, authored by Shivani Mohan, VP of Data Science & UXR for Google Search. The post covers AI Mode's US launch in May 2025 through April 2026 and reports that follow-up queries grew more than 40% on average per month, image-input queries also grew more than 40% month over month, brainstorming-style queries grew 30% faster than the overall AI Mode query pace, and planning-style queries grew 80% faster. Search Engine Journal wrote up the report on July 2, 2026, arguing it confirms a shift from short keyword typing to longer, conversational, personal-context queries.

What did Google actually measure?

Google measured its own AI Mode usage logs and reported category-level growth rates, not raw query volumes, response accuracy, or citation behavior.

The report groups queries into five buckets — Explore, Decide, Learn, Create, and Do — and gives growth rates for some of them (planning up 80% faster, brainstorming up 30% faster) without publishing the underlying totals. That framing tells you AI Mode usage is diversifying past simple lookups. It tells you nothing about how AI Mode selects or credits the pages it draws answers from, which is the part publishers and content teams actually need to plan around.

Why should you be sceptical of these particular numbers?

Because Google defines the categories, picks the comparison baseline, and controls whether the underlying counts are ever published for independent checking.

"Follow-up queries grew more than 40% per month" sounds precise, but a percentage-growth figure with no base number can describe almost any real-world scale. This site has already flagged that Google itself admitted Search Console reporting for AI Search is inadequate — the same company now asking the industry to trust an unaudited behavioral narrative published on its own marketing blog. A vendor that won't show its citation data in Search Console has little standing to be taken at face value on query-length trends either.

What does this mean for how you write for AI Search?

Content built for three-word keywords increasingly misses conversational, context-heavy queries that Google says now make up a growing share of AI Mode traffic.

The practical risk isn't just stale keyword targeting — it's the temptation to solve it by mass-producing "conversational" answer pages with AI writing tools, fast, to match the new query shape. Google's report says nothing about how it treats machine-generated pages competing for those longer queries, and neither Search Console nor AI Mode currently discloses that. If your team is about to brief writers (or a model) to sound more like a person "narrating context into the search bar," ask your own vendor or agency how they're verifying that output isn't detectable as templated AI filler before it goes live.

Frequently asked questions

Is this report based on published, independently verifiable data?

No. It's a Google blog post with percentage-growth claims and no raw query counts or methodology disclosed for outside verification.

Does the report show which pages AI Mode is citing more often?

No. It describes query behavior only — length, phrasing, and category growth — not citation or ranking data.

Should I rewrite all my content around longer conversational queries now?

Audit your top pages against how a real person would phrase the question first; wholesale rewrites based on one vendor's growth percentages are premature.

Source: Search Engine Journal, "Google Data Shows AI Search Users Moved Past Keywords, Your Content Hasn't".