Google reveals first AI Mode usage data after one year

What happened

Per Search Engine Journal, Google has released its first batch of first-party usage data for AI Mode in the United States, one year after launch. The headline figures: AI Mode now has 75 million daily active users in the US, queries inside AI Mode are two to three times longer than traditional Search queries, and roughly 80% of users say they get value from a single AI Mode query without needing a follow-up search.

Google also confirmed that the most common AI Mode use cases cluster around how-to questions, product research, and local exploration. The company framed the disclosure as evidence that AI Mode is additive to Search, not cannibalising it.

Read that framing carefully. Google is not saying AI Mode replaces the ten blue links. It is saying users are now running two distinct query types in parallel: short navigational queries in classic Search, and long exploratory queries in AI Mode. Both surfaces matter, and they reward different content.

Why it matters for your brand

The two-to-three-times query length figure is the one to fixate on. It tells you exactly what kind of content AI Mode is pulling from. Short keyword pages built for 2015-era SEO will not surface here. AI Mode is matching long, intent-rich prompts ("what's the difference between a parametric and indemnity-based catastrophe bond for a sovereign issuer") to content that answers the full question in context. If your asset library is still optimised for three-word head terms, you are invisible on this surface.

For financial services brands, this changes the economics of thought-leadership content. A research note that previously had to compete with Bloomberg and the FT for a click now competes to be the cited source inside a generated answer. The winners will be institutions that publish structured, claim-dense explainers on narrow topics: rate transmission, Basel III endgame, private credit underwriting. Asset managers and reinsurers who treat their PDFs as gated lead magnets will get cited less than mid-tier rivals who publish in HTML with clear headings and numbered claims.

For multilaterals and UN-system organisations, the implications are sharper. AI Mode queries skew toward "explain this to me" intent. That is exactly the territory where UNDRR, the World Bank, WHO and the IMF have an authority advantage over commercial publishers. But that advantage only converts into citations if the underlying content is machine-parseable. A 90-page flagship report locked in a PDF behind a download wall does not get cited. The same content rebuilt as a series of structured web explainers does. Communications leads at multilaterals should treat their flagship publications as raw material, not as the final product.

For major industrial groups (Holcim, Siemens, ArcelorMittal and peers), the 80% "single-query satisfaction" figure is the warning. Buyers researching low-carbon cement, grid-scale storage or industrial automation will increasingly form a view from one AI Mode answer. If your competitor's sustainability disclosures, technical specs and case studies are easier for the model to retrieve and synthesise, the buyer's shortlist forms without you. The B2B procurement funnel is being compressed at the awareness stage, and the compression happens inside the model.

Philanthropic and policy institutions face a related problem: framing power. AI Mode synthesises across sources. If the Gates Foundation, Rockefeller, or a major think tank wants its framing of a policy issue (financial inclusion, pandemic preparedness, climate adaptation) to be the one the model defaults to, it needs more than one well-cited report. It needs a corpus: multiple linked pieces, consistent terminology, machine-readable data tables. Whoever has the densest, most internally consistent body of work on a topic gets pulled into the answer.

The signal in context

This is the first time Google has put real numbers behind AI Mode, and the disclosure is strategic. Google is signalling to publishers, advertisers and regulators that AI Mode is a mainstream product (75 million DAUs in the US is roughly the scale of LinkedIn's US user base) and that it is changing query behaviour in ways Search alone cannot serve. Expect more first-party data from Google over the next two quarters as it prepares advertisers for AI Mode monetisation.

The broader pattern across the AI search landscape is now clear. ChatGPT, Perplexity, Claude and Google AI Mode are all rewarding the same content shape: long-form, claim-dense, structured, and published on domains the model already trusts. The brands that built distribution moats around short-tail SEO and paid media are watching those moats drain. The brands that invested early in deep topical authority, published openly, and structured their content for machine retrieval are compounding citations across every major LLM surface. Google's year-one data is not a new trend. It is confirmation that the trend has reached the largest search surface on the internet.