GEO Statistics for Planning AI Search Investment

The verdict from these generative engine optimization statistics is that AI search already affects discovery, citations and clicks, especially for complex questions. The evidence supports funding GEO as an extension of technical SEO and content—not replacing those programs or assigning budget from citation counts alone.

Choose the decision you are making and the evidence type you trust; the table will narrow the relevant figures.

Google says a page must be indexed and eligible for a Search snippet before it can appear as a supporting link in AI Overviews or AI Mode. There are no additional technical requirements for those features beyond established search fundamentals (Google Search Central).

GEO Research Shows Visibility Gains, Not Guaranteed Revenue

1. GEO techniques increased content visibility by up to 40% in an academic benchmark.

The 2024 paper that formalized “generative engine optimization” tested methods across a large benchmark of queries and reported gains of up to 40%. Results varied by subject area. The figure does not promise that editing a commercial page will produce a 40% lift—or any lift—in ChatGPT, Google or another live product (Princeton University’s record of the KDD paper).

2. Only 12% of URLs cited by four AI assistants also ranked in Google’s top 10 for the original prompt.

Ahrefs compared citations from ChatGPT, Gemini, Copilot and Perplexity with search results for 15,000 long-tail queries. Perplexity was the outlier: 28.6% of its citations ranked in Google’s top 10, while the other assistants were near 8% or below. A rank tracker for the literal prompt is therefore not a complete AI-visibility measurement system (Ahrefs).

3. For Google AI Overviews, 37.1% of cited URLs ranked in the organic top 10 for the same query.

In a later analysis of 863,000 SERPs and four million AI Overview URLs, Ahrefs found that another 26.2% ranked in positions 11–100. The remaining 36.7% did not appear in the organic top 100. The overlap was stronger than for the standalone assistants, but it was still far from a one-to-one relationship between rankings and citations (Ahrefs).

Keep technical SEO, indexability and conventional rankings in the plan, but measure AI answers separately. Use a stable set of buyer questions to track cited URLs, brand mentions, competitors and answer accuracy. A citation is visibility, not a lead.

AI Summaries Appear More Often on Complex Queries

Pew Research Center analyzed browsing data from 900 U.S. adults in March 2025. Its findings provide observed behavioral evidence rather than estimates from a keyword database.

4. AI summaries appeared on 18% of the Google searches observed.

5. Fifty-eight percent of participants encountered at least one Google search with an AI summary during the month.

6. Fifty-three percent of searches containing 10 or more words produced an AI summary.

That compares with 8% of searches containing only one or two words.

7. Sixty percent of searches beginning with a question word produced an AI summary.

The sample is U.S.-based and covers one month. It nevertheless shows that longer, question-shaped searches—the kind often addressed by research and educational content—were especially likely to generate summaries.

AI Summaries Correlate With Fewer Search Clicks

The same Pew study measured what users did after encountering the results page.

8. Users clicked a conventional result on 8% of visits with an AI summary, versus 15% of visits without one.

9. Users clicked a source inside the AI summary on 1% of visits containing a summary.

10. Twenty-six percent of visits to an AI-summary results page ended the browsing session, compared with 16% of visits to a conventional results page.

All seven Pew figures come from the same Pew Research Center study. They show why traffic forecasts based only on rank and search volume are becoming less dependable.

11. Ahrefs estimated that an AI Overview correlated with a 58% lower average click-through rate for the position-one page.

The analysis compared informational-keyword click-through rates in December 2023 and December 2025, forecasting what the later rate would have been without AI Overviews. It was a modeled, observational comparison rather than a randomized test, so “correlated with” is the appropriate interpretation (Ahrefs).

Forecast fewer visits per impression on affected queries. Report impressions, AI-answer presence, citations and qualified conversions together rather than treating declining click-through rate as proof that content has stopped influencing buyers.

AI Influences Multiple Stages of B2B Buying

A 2026 Semrush survey collected 622 valid responses from U.S. B2B professionals. The detailed purchasing findings below apply to the 519 respondents who said they use AI for work, not to all B2B professionals.

12. Sixty-six percent regularly used AI to research products, vendors or solutions; another 29% did so occasionally.

13. Ninety-two percent said AI had shaped a vendor shortlist, including 45% who described the impact as significant.

14. Seventy-two percent used AI during early category research, 62% for active vendor comparisons and 45% to support a final decision.

15. After an AI tool mentioned a vendor, 71% visited its website and 63% searched for the company on Google.

This behavior supports a combined AI-search, website and SEO program rather than an AI-only acquisition strategy.

16. Only 7% said brand recognition was what made a vendor stand out in an AI response.

Specific use-case fit mattered to 53%, while a clear, detailed description mattered to 50%.

The results and sample details are available in Semrush’s B2B buyer survey. The findings are self-reported and come from a marketing software company, but they point to concrete content needs: specific use cases, direct comparisons, accurate pricing or contract information, implementation details and verifiable outcomes.

AI Referrals Are Growing but Remain Complementary to Search

17. AI platforms generated an estimated 1.1 billion referral visits in June 2025, up 357% year over year.

18. Similarweb estimated that AI referrals to transactional sites converted at roughly 7%.

19. About 95% of ChatGPT users still used Google.

These figures come from Similarweb’s analysis of billions of web and app signals. They are proprietary estimates and extrapolations rather than audited totals, a limitation Similarweb states in its release (Similarweb). Together, they describe a fast-growing channel that remains complementary to search.

Conversion performance can vary sharply by site. In one Seer Interactive client account covering October 2024 through April 2025, ChatGPT referrals converted at 15.9%, compared with 1.76% for Google organic traffic. All AI sources combined, however, produced traffic equal to only 0.07% of organic sessions.

That account illustrates high intent paired with low volume. It is not an industry benchmark that every company should put into a forecast (Seer Interactive).

Crawler Activity Does Not Establish Referral Value

20. Cloudflare observed about 70,900 Anthropic-associated HTML crawl requests for every referral during one week in June 2025.

Cloudflare cautioned that visits from Claude’s native app may omit referral headers, potentially overstating the crawl-to-referral ratio by an unknown amount. It also aggregates relevant user agents at the platform level. Even with those limitations, the figure shows why crawler access and referral value should be evaluated separately rather than assuming that more crawling means more visits (Cloudflare).

A GEO Scorecard Needs Five Measurement Layers

These statistics justify measurement, not an automatic spending level. A workable monthly scorecard has five layers:

1. Eligibility: Indexable pages, permitted crawl access, snippet eligibility and structured data that matches visible content where structured data is used.
2. Visibility: Answer presence, brand mentions, citation share and cited URLs across a fixed prompt set.
3. Accuracy: Correct product capabilities, audience, price, location and limitations in generated answers.
4. Response: Identifiable AI referrals, engaged sessions and assisted journeys from branded search or direct traffic.
5. Business value: Qualified leads, opportunities, pipeline and revenue connected to those journeys.

For a fuller measurement design, use the five metric layers for a post-launch content feedback loop. The budget decision should turn on qualified pipeline and accurate buyer discovery, not a rising citation count by itself.