New research from SparkToro, using Similarweb’s desktop and mobile clickstream panel across US searches from January to April 2026, found that 68.01% of Google searches ended without a click. That is not 68% of searches failing to reach your specific site. It is 68% of all searches ending without a visit anywhere: not to the open web, not to Google’s own properties, not even to a paid result.
For an industry that has spent two decades measuring success in rankings and clicks, this is one of the clearest signals we have seen that measurement needs to evolve alongside the technology. The more useful question now is not only “how do we get more clicks,” but also “what is still earning visibility, and are we measuring it?”
The Numbers Behind the Shift
This trend is not new, but its speed is.
Source: SparkToro / Similarweb, 2026
That’s 7.5 percentage points of growth in just two years, the fastest acceleration in a decade.
The driver is not one single feature:
- AI Overviews now appear on more than 20% of all searches
- When an AI Overview shows, the click-through rate drops by nearly 60%
- An Ahrefs tracker of 75,000+ websites recorded a 22% drop in Google’s traffic share over a single year, even across sites with active SEO management
The takeaway for teams doing the right SEO work is not that the effort is wasted. It is that the goal posts have moved, and where that effort is directed needs to move with them.
A worthwhile caveat: it is not AI Mode, yet
It is tempting to pin this entirely on conversational AI search. The data does not support that. AI Mode accounted for only 0.34% of searches between January and April 2026, despite surpassing one billion monthly users and more than doubling its query volume each quarter. What got Google to 68% is a decade of incremental UI and feature changes designed to keep people inside Google’s own ecosystem, not one dramatic AI launch. The mechanics of that shift are covered in this blog: Google’s AI-Driven Search Overhaul.
Who Is Still Winning the Clicks That Remain
If two-thirds of searches stay on Google, the other third is still up for grabs. Cyrus Shepard at Zyppy ran his own analysis of more than 400 winning and losing websites and identified the traits that consistently separated them.
What winning sites have in common:
- Offer a product or service
- Own proprietary assets
- Stay tightly focused on a niche
- Carry a recognisable brand
- Let the user complete a task on the page itself
Nearly 84% of winning sites let visitors finish what they came to do, compared with about half of the sites that lost ground. The sites losing ground tended to be informational in nature: content that explains something well but then routes the user elsewhere to act on it.
These traits stack. One strength is not enough. The advantage comes from being a genuine destination where the user can finish the job.
The opportunity for service-based sites
This is actually a useful finding for service businesses. Pages that explain a procedure or product clearly, and also give the visitor a direct path to act (book, get a quote, use a calculator, contact a team), are the pages now holding their ground. If a page is already performing well on content depth, adding a stronger on-page action step is a relatively low-lift way to strengthen its position. Booking widgets, calculators, quote tools, and clear next-step CTAs are not just conversion features anymore. They are signals that a page belongs in the winning column.
AI-Sourced Traffic Converts Better, Even Though There Is Less of It
There is a real tension in the data worth sitting with. Clicks are falling, but the clicks arriving through AI sources are increasingly valuable.
| Metric | Figure |
|---|---|
| YoY growth in AI-sourced retail traffic | ~400% |
| Conversion lift vs. non-AI traffic | ~42% better |
| US shoppers who’ve used AI to research or buy | ~3 in 4 |
| Shoppers who won’t let AI spend autonomously | ~1 in 3 |
| Source: Adobe Digital Insights Q2 2026; Exploding Topics 2026 | |
People use AI for the early, exploratory parts of the journey: comparing brands, narrowing options, and summarising reviews. By the time a human follows through to a click, they are already further down the funnel and more ready to act. This is consistent with the “traffic down, revenue up” pattern reported by brands including HubSpot, and it directly supports the measurement shift we wrote about in AI Search Breaks Last-Touch Attribution.
What Still Earns a Click, and What to Measure Instead
1. Brand is the strongest long-term lever
Strong-brand sites won at roughly twice the rate of everyone else in Shepard’s data. According to SparkToro’s own query research, branded search terms make up around 44% of all Google searches. AI systems also default to citing entities they already recognise, a pattern we explored in AI Overviews Grew by 52% in One Year. The strategic priority is to be a named, recognised entity, not just a page that ranks.
2. Build measurement around correlation, not last-click attribution
Organic traffic remains a meaningful indicator of SEO health, and it is not going away as a reporting metric. What is shifting is that traffic alone no longer tells the full story. A correlation dashboard that sits alongside traffic reporting, tracking branded search volume, direct visits, share of voice, and revenue movement gives a more complete read on what is actually working. This is particularly relevant as last-click attribution becomes less reliable due to AI-influenced journeys. We go deeper into this in AI Search Breaks Last-Touch Attribution.
3. Keep publishing, even where clicks will not follow
Pages that earn few direct clicks still carry real weight, because AI systems draw from the public record before deciding what to cite or recommend. A clear, accurate, well-structured page shapes what appears inside an AI Overview or chatbot answer, even when the person reading that answer never visits the site. Publishing consistently is how brands stay in the pool that AI systems pull from.
4. Treat owned channels as the only ground that cannot be taken away
Email lists, communities, and first-party data are not affected by algorithm updates. Distributing content across platforms builds awareness and feeds the AI citation pool, but the audience you actually own is the one that compounds over time, regardless of what any platform changes next.
Why We Care
Ranking first still matters. Organic authority influences AI Overview inclusion, and pages that rank well are cited in AI responses at significantly higher rates than those that do not. That investment is not wasted.
What this data signals is that strong rankings and strong traffic growth are no longer always the same outcome, and that a site can be performing well in search even when raw visit numbers are flat. The opportunity is in expanding what we look at alongside traffic: branded search growth, citation presence in AI answers, and the quality of the visits that do convert.
The brands building that fuller picture now are the ones who will be able to point to real commercial impact from their SEO, even as the click landscape keeps evolving. That is what a good SEO strategy looks like in 2026.