Search Is Sending Fewer Clicks. Here Is the 2026 Data, and What to Measure Instead.
Direct answer: Yes, your organic clicks are probably down, and no, it is not a penalty. In a 2026 randomized field experiment, Google AI Overviews cut organic clicks by 39.8 percent on the searches where they appeared, and most US Google searches now end with no click at all. The site is not broken. The move that matters is not chasing the lost click. It is changing the scoreboard, so you measure whether you are the answer the engine cites and whether that presence shows up later as branded search and direct visits.
If your traffic reports look worse this year and no one can explain it, this piece explains it, and gives you a different set of numbers to run your marketing against.
What the 2026 data actually shows
Three findings, each from a named source, tell the story.
First, the click loss is measured, not anecdotal. Researchers Saharsh Agarwal and Ananya Sen ran a randomized field experiment and found a 39.8 percent decrease in organic clicks on searches where Google's AI Overview was shown. AI Overviews were triggered on roughly 41 percent of the queries in the study. An earlier version of the same paper reported a 38 percent drop, so the effect grew as they refined it. One honest caveat: this is a working paper on SSRN that has not completed peer review.
Second, the shift toward no click is now the norm. From January to April 2026, 68.01 percent of US Google searches ended without a click, up from 60.45 percent in 2024. Separately, Ahrefs has measured that when an AI Overview is present, click-through rate on the top organic result drops by about 60 percent.
Third, the lost clicks were not junk. Google's public explanation has been that AI Overviews mostly absorb low-value bounce clicks. The same experiment tested that claim and found no measurable difference in bounce rate, time on site, or return-to-search between the clicks people made with and without the summary present. The authors wrote that the result is at odds with the view that AI Overviews primarily eliminate low-engagement website visits.
Put together: the engine is answering more questions itself, on a large and growing share of searches, and the traffic it removes is ordinary traffic, not filler.
Why this is not a penalty, and why that matters
When traffic falls, the first instinct is to assume something is wrong on the site. A technical error, a ranking drop, a Google penalty. In most cases in 2026, none of that is the cause.
The experiment found the losses concentrated in informational queries, the "how does X work" and "what is Y" searches where an AI summary can answer in place. Navigational and transactional queries, the ones closer to a purchase, showed no measurable change. That pattern is a diagnosis, not a verdict on your website. Your top-of-funnel explainer content is being read inside the answer instead of on your page.
The implication for a small team is direct. If you spend the next quarter trying to recover that informational traffic with more of the same content, you are fighting the format. The click is not returning to the old number. The value has to be captured elsewhere in the journey.
The upside almost no one measures
Here is the part that reframes the problem. Being present in AI answers appears to drive real behavior, it just does not show up as a click on the answer.
Similarweb studied the downstream impact of AI visibility across finance, travel, and beauty, using clickstream panel data that follows real devices over time. SparkToro's Rand Fishkin, who contributed to the study, summarized the finding: when an AI tool recommended a brand, devices that saw that recommendation were measurably more likely to visit the brand's site within the following seven days. Visitors were 7.2 percent more likely to visit American Express and 14.2 percent more likely to visit Capital One after those brands appeared in AI answers. The channel mix of those later visits shifted too, with search slightly lower and direct visits higher.
The honest caveat, which Fishkin makes himself: this is aggregate data on large, well-known brands, and it leaves open whether the same effect holds for smaller or unknown brands. Treat it as directional evidence, not a guarantee. But the direction is the point. AI presence is not a vanity metric that dead-ends. It leaks into direct traffic and branded search, which are numbers most owners already trust.
The scoreboard I use instead
The old dashboard was built for a world where discovery ended in a click. Most of it still works. It just cannot be the only thing you watch. This is the swap I make with clients.
- Organic sessions from Google becomes organic sessions plus AI-feature impressions. Reason: the click is one outcome of visibility, no longer the only one. - Keyword rankings becomes rankings plus whether you are cited in the AI answer. Reason: position one still matters, but the answer box sits above it. - Total traffic becomes branded search volume and the direct-visit trend. Reason: that is where AI-influenced demand shows up when there is no referral click. - Time on page becomes assisted conversions and lead quality. Reason: fewer, higher-intent visits beat more low-intent ones.
None of this requires new paid tooling to start. Google Search Console now has generative AI performance reports that show impressions for pages appearing in AI Overviews and AI Mode, with dimensions including pages, countries, and devices. That is a free, first-party place to see whether your content is showing up inside the answer, even when it does not earn the click.
What to do in the next 30 days
Three moves, rated for effort and impact so you can sequence them.
First, re-baseline the report before you react. Split your search traffic by query intent and stop treating an informational-traffic dip as an emergency. Add branded search and direct visits to the same dashboard so you can see demand that arrives without a referral. Effort: Low. Impact: High.
Second, make your best evidence quotable. AI answers pull and cite clean, specific claims far more easily than they summarize a long narrative. Put your proof, the named result and the before-and-after number, on a page a machine can lift in one line. This is the same reason a proof page beats a buried case study. Effort: Medium. Impact: High.
Third, turn on the free AI-visibility view. Open the Search Console generative AI performance report and start a monthly note on which pages appear in AI features. You cannot manage presence you are not watching, and this costs nothing but the setup. Effort: Low. Impact: Medium.
The through-line: discovery now happens before the click, and often instead of it. The businesses that hold steady are the ones that stopped grading themselves only on traffic and started grading themselves on being the answer.
Frequently Asked Questions
Q: Is my traffic drop a Google penalty? A: Almost certainly not. In the 2026 experiment, click losses were concentrated in informational queries where an AI Overview answered in place, not in penalized or broken pages. Check whether your dip is in informational content before assuming a technical fault.
Q: Should I block AI Overviews or exclude my site from AI features? A: Usually no. Google Search Console offers a setting to exclude your content from AI Overviews and AI Mode, but excluded content simply will not appear in those features, and the setting is not used as a ranking signal for classic results. For most small businesses, presence in the answer is the goal, not absence from it.
Q: If clicks are falling, is SEO dead? A: No. Search still sends clicks for branded, local, and high-intent transactional queries, which are the searches closest to a sale. What has changed is that informational discovery increasingly happens inside the answer, so the goal shifts from earning every click to being the cited source.
Q: How do I know if I am showing up in AI answers? A: Start with the free Google Search Console generative AI performance reports, which show impressions for pages appearing in AI Overviews and AI Mode. Paid tools can track citations across multiple engines, but the Search Console view is the no-cost place to begin.
Q: What is one metric to add this month? A: Branded search volume. When AI answers influence someone, the follow-up often looks like a direct visit or a branded search rather than a click on the answer, so branded search is a practical early signal that your visibility is working.