Google says AI features send 'billions of clicks' a week. What that does and doesn't tell you

What Alphabet said on 22 July

Alphabet's chief executive gave three numbers about AI in Search in the second-quarter 2026 earnings remarks published on 22 July 2026. AI Mode has passed 1 billion monthly active users since its global expansion in October 2025. The Gemini app has 950 million monthly active users. And Google says it is "now sending billions of clicks to websites every week through AI features in Search."¹

The same remarks say AI is driving growth in queries and report 17% revenue growth in Search and other.¹ These are company statements made to investors, not audited traffic data. They are still worth reading closely, because the clicks line will be quoted in every agency deck for the rest of the year.

Our view: the user counts are useful context and the click claim is not evidence about your site. Here is why, and what to measure instead.

Why a platform total can't answer your question

A weekly click total says nothing about the rate per search, and the rate is what a single site lives on. "Billions of clicks" sounds large, but Google handles a very large number of searches each week. Without the denominator, the sentence cannot be turned into a click-through rate.

Three other things are missing. There is no baseline, so we cannot tell whether AI features send more or fewer clicks than the classic results they sit above. There is no definition of a click to "websites", so we cannot tell how much goes to large platforms, forums or Google's own properties. And there is no breakdown by query type, which matters because AI features appear far more often on some queries than others.

None of this makes the claim false. It makes it a statement about Google's ecosystem, which is not the same object as your traffic.

What site-level studies measure

Studies that look at individual queries and pages tell a different, narrower story. They measure the click rate on a given result, with and without an AI Overview above it.

Ahrefs, which sells SEO software, compared 300,000 keywords on desktop in December 2025 against December 2023 (geography not stated). It found the top-ranking page's average click-through rate was 58% lower when an AI Overview appeared.² SparkToro, using Similarweb's US clickstream panel, found 68.01% of US Google searches ended without a click in January to April 2026, against 60.45% in 2024 on a different panel.³

Being cited changes the picture. Seer Interactive, a search agency, tracked 53 brands from January 2025 to February 2026. Brands cited in an AI Overview earned about 120% more organic clicks per impression than when they were not cited.⁴ Seer also found organic click-through rate on AI Overview queries recovered from 1.3% in December 2025 to 2.4% in February 2026, though cited pages still trailed queries with no AI Overview by 38% on informational searches.⁴

Site-level evidence

The rate per search tells a different story

58%lower CTR for the top page when an AI Overview appearsAhrefs, Dec 2025 vs Dec 2023 ²

+120%organic clicks per impression when cited in the AI OverviewSeer Interactive, Jan 2025–Feb 2026 ⁴

68.01%of US Google searches ended without a clickSparkToro, Jan–Apr 2026 ³

How both stories can be true

Google's total and the site-level rates do not contradict each other. If query volume grows and AI features appear on more searches, total clicks can stay large while the share of searches that end in a click falls. The clicks that remain also concentrate on the pages the AI feature cites.

That has a practical consequence. The average site and the cited site are having different years. A page that is cited may hold or grow its clicks per impression. A page ranking just below an AI Overview it is not part of may lose a large share. Google's statement averages across both, and across every site on the web.

Our view: treat "billions of clicks" as a reason not to panic about Google as a channel, and nothing more. The decision about your content depends on which of those two groups your important queries fall into.

What should you measure on your own site?

Measure the click rate on your own queries, split by whether an AI feature appears and whether you are cited in it. Four steps get you there.

1. In Search Console, export impressions and clicks for your top commercial and informational queries. Search Console does not report AI Overview impressions separately, so check the queries by hand or with a tracker.
2. For each query, record whether an AI Overview or AI Mode answer appears and whether your page is cited. Run each check on three separate days, because the answers vary.
3. Compare clicks per impression across three groups: no AI feature, AI feature with you cited, AI feature without you.
4. Repeat monthly. The trend in the third group is the one that tells you where to work.

Keep AI Overviews and AI Mode apart when you do this; our explainer on tracking them as two channels sets out why. For how these numbers belong in a board report next to AI referrals, see which AI search number belongs in a board report.

At Sigzen AI, our AI answer monitor runs the same split across six engines: ChatGPT, Gemini, Perplexity, Claude, Copilot and Google AI. It uses official APIs where they are offered and a licensed AI-answer data provider otherwise. The AI Visibility Audit runs each prompt on three separate days, so a single lucky or unlucky answer does not set your baseline.

What the remarks do tell you

The user numbers are the part worth keeping. AI Mode passing 1 billion monthly active users means AI answers in Google are a mainstream surface, not a test.¹ A brand that only tracks classic rankings is now measuring part of what a buyer sees.

Our view: the next time a vendor quotes the clicks line as proof that AI search is good for your traffic, ask for the rate on your queries. If they cannot produce it, they are selling Google's average, not your result.