GPT Insights GEO Wiki
GEO WIKI · USERS & USAGE

AI Overviews: Usage and Reach 2026

Reach, rollout, share of affected searches and click behavior, for the United States and worldwide.

By Hanns Kronenberg · Last reviewed: August 8, 2026 · First published: August 8, 2026

AI Overviews 2026: state of play

Worldwide reach (official Google figures) and the U.S. market.

Global
> 2.5 bn
Monthly active users
„monthly active users"·global·May 2026 (I/O)
Official
Source: Google (I/O 2026)
200+
Countries & territories
„countries and territories"·global·May 2025
Official
Source: Google
40+
Languages
„languages"·global·May 2025
Official
Source: Google
United States
≈ 16–18 %
Share of U.S. searches with an AI Overview
search exposure (Semrush keywords / Pew real sessions)·United States·2025
Market measurement
Search exposure, not the same as user reach. Studies vary by sample and definition.
Source: Semrush · Pew Research Center
Reach ≠ search exposure

Two entirely different quantities: reach = how many people use AI Overviews (> 2.5 bn monthly active users, global). Search exposure = the share of searches where an AI Overview actually appears (e.g. ≈ 16–18 % in the U.S.). The two are never merged into a single metric.

AI Overviews ≠ AI Mode

AI Overviews is the AI answer block within classic Google Search. AI Mode is a separate, conversational AI search mode (per Google > 1 bn monthly active users, I/O 2026) with its own core metric. Both live inside Search but are kept strictly separate here. AI Mode gets its own page.

Quick definition · continuously maintained reference (Users & Usage)

This continuously updated reference documents the reach and usage of Google AI Overviews (AIO), the AI answer block inside Google Search, worldwide and in the United States. Reach (users, rollout) and search exposure (share of affected searches) are kept strictly separate; in-house SERP measurements are labelled as such.

How widespread are AI Overviews in the United States?

Share of affected searches and the click effect, as far as reliably measured.

≈ 16 %
Share of searches with an AI Overview (keywords)
AIO share of queries (10 m+ keywords, US desktop)·United States·Nov 2025
Market measurement
Fluctuated widely through 2025 (6 % → ~25 % → ~16 %).
Source: Semrush
≈ 18 %
Share of real searches with an AI summary
share of searches with AI summary (browser tracking, people panel)·United States·Mar 2025
Representative
Source: Pew Research Center
8 % vs. 15 %
Click on a result link
with vs. without an AI summary·United States·Mar 2025
Representative
Click on a link inside the AI Overview: only ~1 %. Sessions ended more often afterwards (26 % vs. 16 %).
Source: Pew Research Center

AI Overviews reach since 2024

Milestones with the exact metric wording. Google changes the phrasing over time.

  1. Oct 2024
    > 1 bn Official
    „users on a monthly basis"·global
    Source: Google (Q3 2024 earnings)
  2. May 2025
    ~ 1.5 bn Official
    „monthly users"·global
    Source: Google (I/O 2025 official recap)
  3. Jul 2025
    > 2 bn Official
    „monthly users"·global
    Source: Google (Q2 2025 earnings)
  4. May 2026
    > 2.5 bn Official
    „monthly active users"·global
    Metric wording shifted: 2024 „users on a monthly basis", 2025 „monthly users", 2026 explicitly „monthly active users".
    Source: Google (I/O 2026)

AI Overviews search exposure: share of affected searches

How often does an AI Overview appear at all? Market / SERP measurements, kept strictly per market.

≈ 16 %
United States (keyword coverage)
AIO share of queries (10 m+ keywords, US desktop)·United States·Nov 2025
Market measurement
Source: Semrush
≈ 18 %
United States (real sessions)
share of searches with AI summary (browser tracking)·United States·Mar 2025
Representative
Source: Pew Research Center
≈ 20 %
Germany (comparison)
AIO share of queries (100 m+ keywords)·Germany·Feb 2026
Market measurement
International comparison, not a U.S. figure.
Source: SISTRIX
Why the numbers vary

Exposure studies range from ~16 % to over 50 % depending on keyword sample, country, device, timing and definition. They are not directly comparable and are not merged here. A market measurement (keyword/SERP database, e.g. SISTRIX/Semrush) ≠ a representative people study (e.g. Pew, browser tracking).

Click behavior with AI Overviews

Does an AI Overview change click behavior? Real measurements, kept per market.

8 % vs. 15 %
United States, click on a result link
with vs. without an AI summary (real sessions)·United States·Mar 2025
Representative
Click on a link inside the AI Overview: ~1 %. Sessions ended more often (26 % vs. 16 %).
Source: Pew Research Center
27 % → 11 %
Germany. CTR position 1 (organic)
when an AIO is present (−59 %)·Germany·Feb 2026
Market measurement
Source: SISTRIX

Own SERP measurement: recipe galleries give way to AI / knowledge containers

A GPT Insights observation across 1,000 mobile U.S. food keywords, clearly separate from the Google and third-party figures.

GPT Insights tracks the structural change of Google's results page on a fixed set of 1,000 mobile U.S. food keywords (Recipe Gallery Monitor). This does not measure Google's reach but the observable shift of the SERP in a clearly defined topic area: from classic recipe galleries toward AI/knowledge containers.

94 % → 45 %
Recipe galleries (mobile, U.S. food)
share of keywords showing a recipe gallery·United States·Feb 2026 → now
Own SERP measurement
Source: GPT Insights. Recipe Gallery Monitor
487
Lost their gallery
keywords (of 1,000) since Feb 2026·United States·2026
Own SERP measurement
Source: GPT Insights. Recipe Gallery Monitor
76 %
of those, showing a knowledge-graph / entity block
of the gallery losses·United States·2026
Own SERP measurement
Source: GPT Insights. Recipe Gallery Monitor
30.7 %
Classically detected AI Overviews
detected as a standalone AIO·United States·2026
Own SERP measurement
Source: GPT Insights. Recipe Gallery Monitor
≈ 68 %
Total detectable AI presence
incl. condensed AI Overviews inside the entity block·United States·2026
Own SERP measurement
Do NOT read this as „AI Overviews 68 %", it is the broader category of detectable AI presence.
Source: GPT Insights. Recipe Gallery Monitor
Key finding

In this U.S. food set, classic AIO detectors capture possibly less than half of the actually visible AI presence (30.7 % detected vs. ≈ 68 % total). The reason: generative elements partly appear inside a knowledge-graph / entity container and are therefore not counted as a standalone AI Overview.

Scope & limits

Basis: 1,000 mobile U.S. food keywords. The results are an in-house SERP measurement and not a general figure for Google Search as a whole. Read as a U.S. leading indicator, not directly transferable to Germany.

Methodology and evidence levels

Metric hygiene

Three quantities are kept strictly separate: reach (users/rollout), search exposure (share of affected searches) and click behavior (CTR effect). Also: AI Overviews ≠ AI Mode. Google's wording changes: „users on a monthly basis" (2024) → „monthly users" (2025) → „monthly active users" (2026), taken verbatim from each source, never flattened.

Every metric carries an evidence level, so it stays visible whether a figure comes officially from Google, from a large market/SERP measurement, from a representative people study, from an in-house SERP measurement, or from an estimate.

Official
A metric published directly by Google (earnings call, I/O, blog.google).
Market measurement
A large third-party market/SERP measurement across very many keywords (e.g. SISTRIX, Semrush). Very extensive, but not a representative sample of people.
Representative
A representative population/user study (e.g. Pew, YouGov, Bitkom).
Own SERP measurement
A GPT Insights observation (Recipe Gallery Monitor). Kept clearly separate from Google and third-party figures.
Estimate / Context
Modelling/extrapolation as a range, or supplementary orientation, not a direct measurement.

Historical AI Overviews reach and revisions

Reach over time, with the metric wording used at the time.

Datentabelle: As of · Value · Metric wording / context
As ofValueMetric wording / context
Oct 2024> 1 bn„users on a monthly basis" (Q3 2024 earnings)
May 2025~ 1.5 bn„monthly users" (I/O 2025 official recap)
Jul 2025> 2 bn„monthly users" (Q2 2025 earnings)
May 2026> 2.5 bn„monthly active users" (I/O 2026), metric wording changed

Changelog

  • August 2026 (first publication): reach time series 1 → 2.5 bn (Google earnings/I/O, exact metric wording: „monthly users" → „monthly active users"); rollout 200+ countries / 40+ languages; search exposure U.S. ≈ 16–18 % (Semrush/Pew) and Germany ≈ 20 % (SISTRIX); click behavior Pew/SISTRIX; own SERP measurement (1,000 U.S. food keywords) with key findings (94 %→45 % galleries, ≈ 68 % AI presence vs. 30.7 % classically detected).

Sources and further documentation

More in the GEO Wiki
Crawler Accessibility
How to determine whether search, AI and other web crawlers can actually reach a website's content.
More in the GEO Wiki
User-Agent and WAF-Based Access Restrictions
How user-agent rules, web application firewalls, bot management, challenges and other technical signals affect automated access.
More in the GEO Wiki
ChatGPT Users and Usage 2026
Official global user numbers, growth, usage patterns and a transparent U.S. estimate for ChatGPT usage, clearly separating measured figures from extrapolations.
More in the GEO Wiki
Google AI Mode: Usage and Reach 2026
Reach, query growth, country and language rollout, and the U.S. usage patterns of Google AI Mode, worldwide and for the U.S., kept strictly separate from AI Overviews.
More in the GEO Wiki
Google Gemini: Users and Usage 2026
Reach, growth, usage and model status of the Google Gemini app, worldwide and for the U.S., kept strictly separate from Gemini models/API and Gemini in Search.
More in the GEO Wiki
Claude (Anthropic): Users and Usage 2026
Reach (as far as official) and observed usage of Claude via the Anthropic Economic Index, worldwide and for the U.S., with a clean denominator and three separated levels of claim.

Last reviewed: August 8, 2026