WWikiAP
Category: Data

GEO & AEO Key Statistics 2026 — With Sources

A citation magnet that gathers verifiable statistics on the adoption of AI search, citation, and generative search, each with its source URL. It spans everything from the visibility lift reported in the GEO paper to zero-click rates and AI-summary click-through.

Editorial LeadPublished Updated

Generative engine optimization (GEO) and answer engine optimization (AEO) are awash in intuition and anecdote, but verifiable numbers are scarce. This document is a citation magnet that selects only well-sourced statistics and organizes them as "figure / meaning / source," so you can quote them directly in GEO and AEO decisions. Every number carries a real original URL, and any statistic that could not be verified was deliberately excluded.

Why you should gather figures together with their sources

The problem is simple. AI search statistics float around quoted as a single line — "X% of all searches are zero-click" — but that X differs from study to study. The cause is differences in measurement method. Each study uses a different clickstream panel, keyword sample, device and region mix, and detection algorithm. As a result, even the same metric can diverge by nearly a factor of two depending on the source.

Ignore that gap and the impact spills into practice. Cite a wrong single figure in a proposal or an executive report and the credibility of the strategy collapses — and you end up violating the very point the GEO paper stresses: the more statistics and citations you put in the body, the more you get cited. So the operating principle is clear. Always record a figure together with its source URL and measurement period, and present it as a range rather than a single ground-truth value.

Key statistics table

FigureMeaningSource
Up to 40%Adding citations, statistics, and quotations to the body lifts source visibility in generative-engine responses by up to 40%arXiv 2311.09735
30–40%Improvement range on the Position-Adjusted Word Count metric for the techniques above (varies by metric and domain)arXiv 2311.09735
8% vs 15%The click-through rate to search results on visits with an AI summary (8%) is about half that of visits without one (15%)Pew Research, 2025-07
1%Share of visits that clicked a link inside the AI summaryPew Research, 2025-07
About 18%Share of March 2025 Google searches accompanied by an AI summary (sample of 68,879)Pew Research, 2025-07
68.01%Zero-click rate for U.S. Google searches (January–April 2026, Similarweb clickstream)SparkToro, 2026
60.45%The 2024 value of the zero-click rate above → a 7.56pp rise over two yearsSparkToro, 2026

Seen as charts

Two of the figures from the key statistics table are condensed into charts. Each chart is paired with an extractable data table (because it is text and a table, not an image, AI reads it as-is too).

Click-through rate by presence of an AI summary No AI summary 15% AI summary present 8% Link within summary 1% Source: (Pew Research, 2025)
Click-through rate to search results by presence of an AI summary — Source: (Pew Research, 2025)
ItemValue (%)Source
No AI summary15%(Pew Research, 2025)
AI summary present8%(Pew Research, 2025)
Link within summary1%(Pew Research, 2025)

When an AI summary appears, the click-through rate to search results nearly halves from 15% to 8%, and clicks on links inside the summary stop at just 1% (Pew Research, 2025).[2] The more clicks are absorbed into AI answers, the more "being cited as a source" — rather than link ranking — becomes the core of visibility.

Zero-click rate for U.S. Google searches over time 60.45% 2024 68.01% Jan–Apr 2026 Source: (SparkToro·Similarweb, 2026)
Trend in zero-click rate for U.S. Google searches — Source: (SparkToro·Similarweb, 2026)
ItemValue (%)Source
202460.45%(SparkToro·Similarweb, 2026)
Jan–Apr 202668.01%(SparkToro·Similarweb, 2026)

Over the same span, the zero-click rate for U.S. Google searches rose 7.56pp, from 60.45% to 68.01% (SparkToro·Similarweb, 2026).[3] SparkToro pointed to the spread of AI Overviews as a major accelerant, while noting that it did not attribute the entire increase to AI summaries.

Context and caveats for each statistic

The GEO paper (up to 40%). arXiv 2311.09735 is a result validated by researchers from Princeton, Georgia Tech, AI2, and IIT Delhi using a large-scale benchmark called GEO-bench. Its core message is "the more citations, statistics, and credible quotations you add to content, the more often it is cited by generative engines," and the size of the lift varies by metric and query domain. So 40% is an upper bound, not an average guaranteed for every page.[1] This very insight is also why this document attaches a source to every figure.

Click behavior (Pew). Pew Research Center analyzed 68,879 Google searches from the actual browsing data of 900 U.S. adults (data: March 2025). Visits accompanied by an AI summary clicked through to search results only 8% of the time, and clicks on links inside the AI summary were just 1%.[2] In other words, AI answers absorb clicks, and "being cited" as a source becomes the core of visibility — this is the starting point of AEO.

Zero-click (SparkToro·Similarweb). In the clickstream analysis by SparkToro and Similarweb, the zero-click rate for U.S. Google searches was 68.01% in January–April 2026, up 7.56pp from 60.45% in 2024. SparkToro identified the spread of AI Overviews as a major accelerant, while explicitly stating that it did not attribute the entire increase to AI summaries.[3] It is more accurate to include this caveat when you cite the figure.

Operating principles when citing

When you carry these figures into proposals and strategy documents, follow three rules. First, pair every figure with its original URL and measurement period. Second, for metrics with large source variance — such as the AI Overviews appearance rate — use a range like "about 18% (Pew) to 48% or more (depending on the tracker)" instead of a single value. Third, do not mix and compare metrics that measure different things — such as the zero-click rate and the AI-summary click-through rate — in one sentence. For deeper use, refer alongside to the GEO recommended companies guide and the shift in generative search behavior documents.

To work directly on structured data and citability, you can also use Google's structured data guide as a reference document.[4]

Wrap-up

GEO and AEO statistics lose credibility the more they are consumed as a "one-line number." The GEO paper's up-to-40% visibility lift, the click-through rate that drops to 8% when an AI summary is present, and the zero-click rate that has reached 68% are each independent pieces of evidence measured in different ways. The point is not the figure itself but the habit of presenting it with its source — and that habit is precisely what makes content citable by AI.

Frequently asked questions

Q.What exactly does the GEO paper's "up to 40%" mean?
It means that when you apply the optimization techniques proposed in the arXiv 2311.09735 study (Princeton, Georgia Tech, AI2, IIT Delhi), source visibility in generative-engine responses improved by as much as 40%. Adding citations, statistics, and credible quotations was especially effective.
Q.Are the zero-click rate and the AI-summary click-through rate the same metric?
No. The zero-click rate (SparkToro, Similarweb) is the share of search sessions in which no link was clicked at all, while the AI-summary click-through rate (Pew) is the share of visits that clicked a search result when an AI summary appeared. They measure different things on different panels, so they cannot be compared directly.
Q.On what share of all searches does AI Overviews appear?
It varies widely by source. The Pew study reports about 18% in March 2025, while other trackers report 20% to 48% or more. Because keyword sets, regions, devices, and detection methods differ, treat this as a range rather than a single number.
Q.Where should I use these statistics?
Cite them as evidence in GEO and AEO strategy documents, proposals, and executive reports. When you cite them, always include the original source URL and the measurement period, just as we do in the body.
Q.Why do the figures vary so much for what looks like the same metric?
Because each study uses a different clickstream panel, keyword sample, measurement period, and detection algorithm. That is why it is more accurate to disclose the source and method together and present a range, rather than citing a single figure.

Sources

  1. [1] ↑GEO: Generative Engine OptimizationarXiv (Princeton·Georgia Tech·AI2·IIT Delhi)
  2. [2] ↑Google users are less likely to click on links when an AI summary appears in the resultsPew Research Center
  3. [3] ↑In 2026, Less than One Third of Google Searches Still Send a ClickSparkToro · Similarweb
  4. [4] ↑Structured data markup introGoogle Search Central

This document was last edited on Jun 26, 2026. WikiAP content is compiled from public primary sources and updated for accuracy.