For every 1,000 U.S. Google searches, 232 clicks reach the open web. That's the number from SparkToro's latest analysis of Similarweb's clickstream panel, covering January through April 2025. In their 2024 report, using Datos data, the equivalent figure was 360. A 36% drop in open-web clicks per search, in roughly a year.
If your pipeline model assumes organic traffic will grow proportionally with search volume, that assumption just got a lot more fragile.
What the data actually shows
Rand Fishkin's report breaks post-search behavior into three buckets: 39% of searches end with no action at all, 29% lead to another Google query, and 32% produce a click. Of those clicks, 66% go to the open web, 27% stay inside Alphabet properties (YouTube, Maps, AI Mode), and 6% go to paid ads.
The click rate fell from 41% to 32% compared with the 2024 dataset. That 9.5-point drop is a 22% decline, the largest shift among the metrics SparkToro tracked. Searches that led to another search rose 7 points. Users aren't leaving Google. They're just... searching again.
Fishkin points to AI Overviews as the main driver, citing Ahrefs data showing AI Overviews appear on more than 20% of searches and CTR drops nearly 60% when one shows up. Separately, Ahrefs' traffic tracker recorded an 8-point decline in Google's share of referral traffic to tracked sites between June 2025 and May 2026.
The Google counterargument (and its gap)
Google VP of Search Liz Reid has said organic click volume is "relatively stable" and that AI Overviews mostly remove "bounce clicks" where users grab a fact and leave. Google hasn't published data supporting either claim.
There's a real distinction here worth understanding. SparkToro measures clicks per search. Reid's statements reference total click volume, which can hold steady if query growth offsets a falling click rate. Both things can be true simultaneously. But for demand gen, the per-search rate is what matters for forecasting. Total volume being "stable" doesn't help if your category's queries are getting answered on-SERP before anyone clicks.
A few caveats on the data itself: the 2024 panel from Datos had a disproportionate share of ad-blocker users, which inflated the zero-click percentage and suppressed paid click counts. Fishkin calls comparing across years "a bit of apples and oranges." The directional trend, though, aligns with Ahrefs, Pew Research Center (which found traditional result clicks dropped from 15% to 8% when AI summaries appeared), and BrightEdge's reporting of clicks declining nearly 30% since May 2024.
What this means for your pipeline model
The practical impact hits B2B SaaS teams in three places.
Traffic forecasts. If you're projecting SEO-sourced MQLs using 2023 or early 2024 click rates, you're probably over-forecasting by 20–35% on informational queries. Branded and high-intent transactional queries appear more resilient (Fishkin credits Cyrus Shepard's analysis here), but top-of-funnel informational content is absorbing the worst of the decline.
Content mix. Bottom-of-funnel content types still drive clicks. Comparison pages, "alternatives" pages, pricing breakdowns. These queries carry buyer intent that AI Overviews handle poorly because the answers are subjective and context-dependent. If your content calendar is still weighted 70/30 toward informational blog posts, that ratio deserves a hard look.
Measurement. Sessions as the primary SEO KPI is increasingly misleading. SERP visibility, branded search volume, and assisted conversions tell a more complete story. The question isn't "did they click?" anymore. It's "did they see us, and did that impression contribute to a downstream conversion?" Harder to measure. More honest.
The experiment to run this week
Hypothesis: If we shift 30% of new content production from informational top-of-funnel to comparison and alternatives pages targeting buyer-intent queries, then demo requests from organic will hold flat or increase even as total organic sessions decline, because click-through on high-intent queries is more resilient to AI Overview compression.
Setup: Pull your top 50 non-branded informational pages by traffic. Check which ones trigger AI Overviews (Ahrefs or manual spot-checks). Flag the ones where traffic dropped more than 25% YoY. Reallocate that production effort toward comparison content mapped to your competitor set.
Success metric: Organic-sourced demo requests (not sessions). Guardrail: Total organic sessions don't drop more than 15% in 60 days. Stop-loss: If demo requests from organic fall 20%+ in 30 days, pause and reassess the content targeting.
When this is wrong: If your product category is so niche that comparison queries have negligible search volume, this trade-off doesn't work. You'd be better off doubling down on branded SERP ownership and distribution through channels that don't depend on Google clicks at all.
Where this goes next
AI Mode accounted for just 0.34% of searches in SparkToro's dataset. Google says usage has passed 1 billion monthly users and queries are more than doubling each quarter. That 0.34% won't stay small.
Fishkin wrote that SEO matters as much as ever; "it just won't earn you traffic the way it once did." That's the right framing. The channel isn't dead. The math underneath it changed. And 232 clicks per 1,000 searches is the new baseline your forecast needs to account for, not the 360 you probably built last year's plan around.
