Geneo Blog
1 min read

Debunking the '94% Zero-Click AI Search' Myth: 2025 Data & SEO Moves

Get the real 2025 AI zero-click stats—under 64%. See how AI Overviews, ChatGPT, and Perplexity impact SEO, plus actionable adaptation tips.

Debunking the '94% Zero-Click AI Search' Myth: 2025 Data & SEO Moves

Updated on 2025-10-07

If you’ve seen the headline that “94% of AI-powered searches are zero-click,” here’s the short answer: the number doesn’t hold up against current, authoritative evidence. The best available data across 2024–2025 points to roughly 58–64% of Google web searches not leading to a click to the open web, with meaningful variance by query type and device. That’s serious—but it’s not 94%. And with AI Overviews, ChatGPT-like experiences, and Perplexity gaining share, marketers face a structural shift: visibility is moving upstream of the click.

This article clarifies what the latest studies actually say, why the misquote matters for your priorities, and how to pivot from classic SEO to Answer/Generative Engine Optimization without resorting to panic or platitudes.

What the data actually says (and what the 94% myth gets wrong)

The most comprehensive recent measurement comes from the SparkToro/Datos analysis of 332 million browser-based Google searches across the US and EU over 21 months in 2023–2024. In that dataset, only 374 out of 1,000 US searches resulted in a click to the open web—implying that about 58–64% did not send a click to an external site, depending on region—and a large share ended the session or reformulated the query. See the details in the SparkToro 2024 post, “For every 1,000 US Google searches, only 374 clicks go to the open web,” which also explains their definitions and exclusions. According to the study in 2024, US zero‑click behavior was 58.5% and EU was 59.7% (SparkToro 2024 Zero‑Click Search Study).

Why the “94%” keeps circulating: cherry-picked screenshots, conflation of “no external click” with “no user action,” and extrapolations from limited cohorts. Precision matters—misstating the scale leads teams to over-rotate on short-term fixes and misread where the real risk sits (informational queries, mobile, and SERPs where AI and other modules consume screen real estate).

2025: AI Overviews, answer engines, and uneven impacts across queries

As generative answers expand, their footprint in search results has grown, especially for informational queries. A July 2025 analysis found that AI Overviews appeared in roughly 6.5% of US desktop queries in January and about 13.1% in March—nearly doubling in two months—with a heavy skew toward informational intent (Semrush 2025 AI Overviews study).

On the publisher side, multiple cohorts reported search referral declines they attributed partly to AI Overviews. Trade-press coverage of publisher data from 2025 shows some DCN members saw Google Search referral drops ranging from low single digits to around 25% as AO exposure grew; at the same time, keywords triggering AO increased significantly for certain outlets (Digiday 2025 publisher referral trends).

One widely discussed user-level study from mid-2025 observed that when an AI summary appeared, users clicked traditional results far less often (about 8% of searches vs. 15% without summaries) and were more likely to abandon browsing afterward. The research sampled roughly 900 US participants, logging 68,879 searches in March 2025, with 12,593 triggering AI summaries (Pew Research 2025 short‑read on AI summaries and clicks). Google disputed the methodology and representativeness, emphasizing that Search still drives massive click-outs and that AI features can stimulate follow-up questions (Google’s response summarized in PPC Land, 2025). Treat these as complementary perspectives: measured behavioral signals pointing to fewer clicks in certain contexts, alongside the platform’s contention that net exploration and total clicks remain robust.

Bottom line for 2025: Impacts are heterogeneous. Informational queries and mobile layouts often show the largest “no-click” effects, while high-intent commercial queries can still perform—especially where clear, trustworthy product/service pages exist.

A practical AEO/GEO playbook for 2025

The goal isn’t just to “rank” anymore—it’s to be cite‑able, quotable, and link‑eligible in AI Overviews and answer engines. Here’s a pragmatic playbook teams can implement within existing workflows:

  • Elevate E‑E‑A‑T and evidence clarity on priority pages

    • Add visible author bios and credentials, bylines, updated-on stamps, and a brief “how we vet data” section.
    • Bind claims to authoritative sources with concise, inline anchors and surface basic metadata (year, sample where relevant).
    • Publish auditable first‑party methods (time windows, samples) when presenting your own data.

  • Structure content for “liftability” and synthesis

    • Present scannable definitions, FAQs, checklists, and step-by-step workflows in neutral language.
    • Use appropriate schema (FAQ, HowTo, Organization, Author) and maintain logical heading hierarchy.
    • Provide canonical signals, minimize JS-only rendering of core content, and prioritize mobile performance.

  • Target the intents that AI answers prefer—and fill gaps

    • Produce up-to-date comparative guides and consensus summaries that aggregate multiple reputable sources.
    • Create distinct pages for high-intent queries; avoid burying answers in long-form content without summary blocks.

  • Measurement-first mindset

    • Track AI citations, linkbacks, and mention frequency across Google AI Overviews, ChatGPT-like experiences, and Perplexity; correlate with changes in organic CTR and referral sessions by query cluster and device.
    • Maintain a living change-log of AI answer presence and screen placements.

Example workflow: monitoring cross-engine AI citations and sentiment

  • Map your top informational clusters and questions that commonly trigger AI answers.
  • Monitor whether your brand/content is cited in those AI results and the sentiment of the reference.
  • Compare presence/position over time and annotate UI or policy changes.

You can operationalize this with a multi‑engine monitoring tool such as Geneo. Disclosure: Geneo is our product. Use it to see where your brand appears across Google AI Overviews, ChatGPT-like experiences, and Perplexity, how often you’re cited, whether links are included, and what the sentiment looks like. Then reconcile that visibility with traditional metrics (organic sessions, assisted conversions) to make prioritization decisions.

Measurement templates and update cadence

A simple measurement framework that works in 2025:

  • Define clusters and intents

    • Group keywords by question intent (what/how/why), device distribution, and funnel stage. Expect higher zero‑click risk in informational clusters.

  • Baseline and annotate

    • For each cluster, capture: AO/answer presence, citation presence/position, links in the summary, and traditional ranking/CTR. Annotate major interface changes and algorithm updates.

  • Track deltas every 4–6 weeks

    • Re-measure presence and click metrics; monitor whether AI summaries are absorbing or amplifying demand. Keep a running log of what changed and what you did in response.

  • Tie visibility to outcomes

    • Where AI citations rise but clicks fall, look for signs of “assisted demand” in brand queries and late‑stage visits. Adjust attribution models to value upstream visibility.

For a concrete, auditable look at cross‑engine visibility patterns, see one of our published query reports analyzing a volatile topic like GPU pricing trends in 2025; it illustrates how citations shift across engines over time and how sources are surfaced (GPU shortage price trends 2025 query report). And when you explore how communities influence what answer engines cite, this deep dive on good Reddit sourcing practices can help you shape credible, cite‑worthy sections on your pages (Reddit communities and AI search citations best practices).

Facts vs. forecasts: what to expect next

  • More UI iteration and volatility: Expect Google to keep adjusting AI Overviews (density, link presentation, citations). Answer engines will experiment with how prominently they display sources.
  • Fewer clicks, higher intent: Average CTR may soften further on informational queries, but the visits that do arrive can be later-stage—and more valuable—if your pages match intent and demonstrate authority.
  • Measurement precedes optimization: Teams that measure AI visibility and sentiment today will move faster, communicate credibly with stakeholders, and avoid overreacting to anecdotes.
  • Cross‑engine readiness is a moat: Content that’s precise, well-cited, and structured travels well—getting cited by Google’s AI Overviews, Perplexity, and chat experiences.

What Google and the ecosystem are signaling

Google’s public guidance continues to emphasize creating “unique, satisfying” content and notes that AI Search surfaces a wider range of sources and links on results pages. The implication: eligibility and clarity matter—make it easy for systems to extract, attribute, and link to your work (Google Search Central guidance, May 2025). Perplexity’s help documentation likewise underscores that answers include numbered citations that link to original sources, reinforcing the case for making your content unambiguous, well‑cited, and technically accessible.

Action checklist you can start this week

  • Add updated-on stamps, author bios, and short methodology blurbs to your top 50 informational pages.
  • Insert concise, neutral summary blocks (definitions, steps, FAQs) and add appropriate schema.
  • Annotate your last three months of AO/answer presence by cluster; note wins/losses and UI shifts.
  • Refresh or create two comparative guides where you can credibly synthesize multiple authoritative sources.
  • Establish a 4–6 week AEO/GEO review cadence that includes a change-log and a brief stakeholder memo.

Sources and further reading (selected 2024–2025)

Change‑log

  • 2025‑10‑07: Debunked the “94% zero‑click AI search” figure; summarized current ranges (~58–64% no external click). Added 2025 evidence on AI Overviews prevalence, heterogeneous impacts, and a practical AEO/GEO measurement workflow.

Closing note If you need a lightweight way to monitor how often your content is cited by AI Overviews and answer engines—and how those mentions trend over time—set up a measurement cadence now. You can start with the workflow above and layer in a specialized monitor when you’re ready. This is the fastest, least speculative path to adapting strategy in a zero‑click era.