Jonathon's AI Wiki

Datos + SparkToro State of Search Q1 2026

9 min read

Source: datos-state-of-search-q1-2026-05-19.md — Eli Goodman (President & Co-Founder, Datos, a Semrush company) / Rand Fishkin (CEO, SparkToro) / Publisher: Datos + SparkToro / Published April 27, 2026 / Data window: March 2025 – March 2026.

Despite relentless hype about AI displacing traditional search, Datos’s Q1 2026 clickstream panel of millions of US and EU/UK desktop users puts AI tools at less than 2% of total desktop web visits — real growth, but from a small base. The report’s more counterintuitive finding is that zero-click search is falling rather than rising: in the US it dropped from 24.5% to 22.4% between December 2025 and March 2026, while organic click share climbed to 44.9%. For practitioners calibrating where to invest in 2026, this is the honest demand-side reality check beneath the tactical AI-citation literature.

Key Takeaways

  • AI tools: under 2% of desktop visits in both the US and EU/UK. Across the full March 2025 – March 2026 panel, AI tools (ChatGPT, Gemini, Perplexity, and others combined) accounted for less than 2% of total desktop web visits in both regions. A parallel Datos “Search Happens Everywhere” analysis covering Q4 2025 put AI tools at approximately 1% of desktop visits vs. traditional search engines at approximately 10%.
  • Zero-click is falling, not rising — US. Zero-click search share in the US fell from 24.5% in December 2025 to 22.4% in March 2026. This is the metric most SEOs feared would spike with AI Overview expansion; the panel data shows the opposite direction.
  • Zero-click is falling, not rising — EU/UK. In EU and UK markets, zero-click dropped from 22.5% to 19.6% over the same window. Both regions moved in the same direction.
  • Organic click share rose to 44.9% in the US. Paid click share held between 2.1% and 2.3% throughout Q1 2026 in the US, above early-2025 levels. EU/UK paid clicks held at 2.3–2.4%.
  • Google engagement at the user level is stable. Searchers were still using Google roughly 100 times per month, and Google’s share of desktop search users held broadly stable across both regions. If a new interface is going to weaken Google, it has to change repeated behavior, not just trial. So far the habit loop is intact.
  • Google AI Mode: fast growth from a small base. The report addresses AI Mode specifically: fast growth, but the absolute numbers remain small relative to traditional search volume. ^[inferred: the report characterizes growth as real but does not publish a specific AI Mode percentage]
  • Search refinement behavior held steady. Users modifying and resubmitting queries (a proxy for search dissatisfaction or exploration) held at 15–16% in the US in Q1 2026, while declining slightly in Europe to around 14% by March.
  • ChatGPT search flattened by late 2025 / early 2026. The “Search Happens Everywhere” companion analysis found ChatGPT’s search activity growth rate flattening in late 2025 and early 2026, even as overall AI tool traffic continued to grow — suggesting the early-adopter surge may be stabilizing.

Methodology

Panel type: Passive clickstream panel of millions of opted-in desktop devices in the United States and the 27 EU member countries plus the United Kingdom. Devices install a lightweight browser extension or ISP-level monitoring; panel members are anonymized and opted in.

What it measures: How real desktop users actually navigate the open web — page visits, search queries, click-throughs, zero-click sessions. This is behavioral truth from actual sessions, not survey recall or estimated traffic.

Data window: March 2025 through March 2026 (12 months). Zero-click figures specifically compared December 2025 vs. March 2026.

Desktop-only caveat: The panel covers desktop web sessions only. Mobile search — which accounts for a substantial majority of total search volume globally — is not represented. Given that mobile-first AI assistants (Siri, Google Assistant, on-device Gemini) and in-app AI usage (ChatGPT iOS/Android) are growing faster than desktop equivalents, the sub-2% AI figure almost certainly overstates AI’s desktop share relative to AI’s total-session share. ^[inferred]

Denominator: Total desktop web visits across the open web — not queries, not unique users. This produces lower AI percentages than query-based analyses because every YouTube video, news article, and e-commerce product page also counts as a visit.

What it cannot see: Mobile search; in-app AI (ChatGPT app, Claude app); voice queries; zero-party behavioral signals inside walled gardens (Gmail, Google Docs, Meta apps).

Comparison caveat: Datos’s zero-click numbers (22–24% range) are not directly comparable to the SparkToro/Fishkin keyword-level zero-click analyses that produced the widely-cited 58.5% figure. The two methodologies answer different questions: Datos counts sessions where the user queried Google and left without a click; SparkToro’s older work estimated queries that resolved entirely within the SERP. Different denominators, different windows.

Where this lands in the AI-SEO cluster

The eight articles in the AI SEO hub collectively advise practitioners on how to optimize content for AI citation. That cluster’s implicit framing — invest more heavily in AI SEO because AI search is becoming a dominant channel — deserves a reality-check anchor, and this is it.

The tactical-layer articles (AirOps fan-out, Zyppy meta-analysis, GEO-16 framework, Digital Applied correlation study) are all correct that AI citations matter and that specific signals raise citation probability. But none of them answer the prior question: how large is the AI search channel right now, and is the zero-click threat as acute as 2024-era projections suggested?

Datos answers both:

  1. AI search is still a thin channel in absolute terms. Sub-2% of desktop visits is a real number that practitioners should carry when setting investment priority. Growing fast from a small base is different from being a dominant channel today. Teams at agencies or in-house SEO functions facing budget allocation decisions need this number to calibrate how aggressively to shift from classical SEO toward AI-citation optimization.

  2. Zero-click is easing, not accelerating. The 2023–2025 SEO industry narrative was that AI Overviews would drive zero-click rates to majority territory, hollowing out organic traffic. The Q1 2026 panel says zero-click is moving in the opposite direction — down roughly 2 percentage points in both regions over one quarter. This suggests the worst of the 2024–2025 AIO disruption may have already passed, and that organic clicks are recovering. ^[inferred: causality is not established; the data shows correlation between AIO stabilization and click recovery, not a direct causal mechanism]

How to use this in planning: Treat it as the demand-side floor beneath the citation-tactics literature. Before implementing any of the AEO/GEO optimization plays from the cluster, confirm the channel size justifies the investment in your specific vertical and traffic mix. For most verticals, the answer in early 2026 is still “classical SEO first, AI-citation investment proportional to current AI channel share in your niche.”

Cross-study disagreement: zero-click methodology

SparkToro (prior analyses, ~2024): zero-click share approximately 58.5% of all queries. Datos Q1 2026: zero-click share 22.4% (US) and 19.6% (EU/UK) as of March 2026. Status: resolved (2026-05-19) — The two figures use incompatible denominators and definitions. SparkToro’s figure counted queries that resolved inside the SERP (including refinements, clicks to ads, clicks to PAA, etc.) as “zero-click” and estimated from keyword-level data. Datos counts desktop sessions where the user ran a Google search and left the browser without clicking any result. Neither is wrong; they measure different things. Practitioners should cite the methodology alongside the number.

Open Questions

  • How fast is the sub-2% figure actually moving? The report says AI tools are growing “fast” but does not publish a quarter-over-quarter growth rate. Without a rate, practitioners cannot model when AI search will reach a threshold that materially changes investment priority. ^[inferred: a specific growth rate number was not published in the public abstract or coverage reviewed]
  • Desktop-only panel — what does mobile add? If AI assistants (on-device Gemini, ChatGPT iOS) are growing faster on mobile than desktop, the sub-2% figure likely understates total AI-attributed session share. A mobile-plus-desktop panel might show a materially higher number.
  • Does the zero-click decline hold across query-intent classes? The Datos panel measures aggregate zero-click across all query types. Informational queries (where AI Overviews are most prevalent) may still show elevated zero-click rates even as the aggregate falls, with transactional queries pulling the average down. Intent-class breakdowns would give practitioners better vertical-specific guidance.
  • ChatGPT search flattening: temporary plateau or structural ceiling? The companion “Search Happens Everywhere” analysis found ChatGPT search activity flattening in late 2025. Whether this represents user-behavior saturation, ChatGPT being displaced by Gemini and Perplexity, or a genuine ceiling on AI search adoption is not answered by this dataset alone.

Try It

  1. Anchor your AI-SEO budget conversation with the channel-size number. Before pitching AEO/GEO optimization to a client or internal stakeholder, lead with the Datos figure: AI tools are under 2% of desktop web visits. Build the investment case on a realistic current share, not on projected-disruption anxiety. The tactics are sound; the urgency calibration matters.
  2. Pull your own zero-click data before assuming you’re under pressure. Datos’s panel shows zero-click falling in Q1 2026. Check your own GSC click-through rates over the same period — if clicks are recovering and impressions are holding, you may be in the same trend. Don’t optimize against a threat that’s currently shrinking in your data. ^[inferred: individual sites will vary; the panel finding is aggregate and directional]
  3. Segment your AI-channel reporting from classical-search reporting. Most Google Analytics / GSC setups do not separate AI Mode clicks from traditional organic clicks. Set up a channel grouping or UTM strategy that lets you track AI-attributed referrals (ChatGPT, Perplexity, Gemini) separately so you build your own share-of-channel data rather than relying solely on third-party panels.
  4. Watch the Q2 2026 Datos update. AI Mode is growing fast from a small base; the Q2 update will show whether the Q1 zero-click decline continues or reverses as AI Mode scales. Set a calendar reminder for approximately July 2026 when Q2 data would normally publish.
  5. Cross-reference with the AirOps fan-out study before doubling down on AEO. The AirOps fan-out study shows that retrieval rank dominates citation probability 4.1× over on-page signals. Datos shows classical search is still 98%+ of the channel. The combined message: invest in ranking well first, invest in AI-citation signals second, and size both investments proportionally to current channel share.

Graph View

Backlinks