Sixty-eight percent of Google searches in the US ended without a single click in the first four months of 2026 — not a click to your site, not to a competitor’s, not even to one of Google’s own properties. That figure comes from a Similarweb clickstream analysis SparkToro’s Rand Fishkin published in June 2026, and it puts real numbers behind something most marketing teams have felt in their traffic dashboards for two years: the click is no longer the default outcome of a search. It is now the exception.

This piece is not going to tell you search engine optimization (SEO) is dead. It isn’t, and teams that treat it that way will lose the queries that still convert. What follows is what actually changed, which parts of your traffic decline are permanent, and what to build instead of a bigger blog.

The last two years erased a decade of stability in search traffic

Zero-click search is not new. SparkToro first measured it in 2016, when 45% of Google searches ended without a click. That rose to 49% by 2019 — the number that originally put the phrase “zero-click search” into the industry’s vocabulary. It reached 60.45% in 2024, based on Datos clickstream data. A separate Semrush analysis put that same year’s US rate at 58.5%, close enough to confirm the trend rather than dispute it. Then, in the four months Fishkin’s team measured at the start of 2026, it hit 68.01%.

That is a 7.56-point jump in two years, on top of eight years of much slower movement. SparkToro calls it the fastest acceleration in the decade it has tracked the metric. The number that should worry a marketing leader more than the headline figure is where that traffic went: “clicks of any kind” — including clicks to Google’s own properties and to paid ads — fell 9.51 points between 2024 and 2026, a 22.9% decline. The share of searchers who simply searched again, without clicking anything, rose 7.2 points over the same period. Google is not losing search volume. It is keeping more of it.

None of this is confined to one clickstream panel. Ahrefs runs an ongoing tracker of more than 75,000 websites whose owners opted in to share traffic data — sites run by people actively fighting for every visitor. Between June 2025 and May 2026, Google’s share of referral traffic to those sites fell 8 percentage points, a 22% drop in a single year.

AI Overviews are the visible driver. They are not the only one.

The obvious explanation is generative AI, and it is a real part of the story. Google’s AI Overviews — the synthesized answer boxes now appearing above organic results — sat on roughly 13% of US queries in March 2025 and now show up on more than 20% of all searches, according to SparkToro’s 2026 estimate. When an AI Overview appears on a page, it cuts click-through rate to the organic results below it by nearly 60%.

But the tidy “AI killed my traffic” narrative has a hole in it. Google’s newer AI Mode — the fully conversational search experience — accounted for just 0.34% of the searches in Fishkin’s 2026 sample, a rounding error next to the scale of the overall shift. Google announced at I/O 2026 that AI Mode had crossed 1 billion monthly users with query volume more than doubling every quarter, so its share will grow. It just hasn’t grown enough yet to explain what already happened. What got Google from 45% zero-click to 68% is two decades of featured snippets, knowledge panels, “People Also Ask” boxes, local packs, and shopping carousels — a walled-garden strategy that predates ChatGPT by a decade and was already accelerating before generative AI joined it. Blaming AI alone lets a marketing team off the hook for adapting to a shift that started long before AI Overviews existed, and one that will keep compounding after the current AI narrative moves on to something else. For the minority of queries that do reach AI Mode specifically, its own zero-click rate is remarkably one-sided: a September 2025 Semrush analysis found it resolves 93% of searches without a single external click.

Not every query lost the same amount of traffic — and that gap is the whole strategy

Averages hide the part that matters. Zero-click growth is not distributed evenly across query types, and the distribution tells you exactly where to stop spending and where to spend more.

Informational queries — definitions, how-to questions, general research — carry a zero-click rate around 74%, per a Semrush intent-based analysis. Transactional queries, the ones closest to a purchase decision, sit closer to 31%. Branded search, which makes up roughly 44% of all Google queries according to the data SparkToro’s Amanda Natividad cites, still reliably sends a click, because a person searching your company name has already decided you are relevant and wants your site specifically.

The pattern is not subtle: the more a query resembles “explain this to me,” the more likely Google answers it without you. The more it resembles “I am ready to act,” the more the click survives. A content calendar built to chase search volume with generic explainer posts is now optimizing for exactly the category of query that stopped sending traffic. A content calendar built around branded, local, and high-intent transactional terms is optimizing for the category that held up.

The sites still winning share five traits, and “better content” isn’t one of them

If two-thirds of clicks disappeared, someone is still earning the third that remains. Cyrus Shepard at Zyppy analyzed more than 400 websites that gained or lost traffic over the same 12-month period and ranked five traits by how strongly each predicted a win. Offering an actual product or service, rather than just information about one, was the single strongest correlate: 70.2% of winning sites sold their own product or service, versus 34.6% of losing sites. Close behind was letting a visitor complete a task on the page itself — book something, calculate something, buy something — true of 83.7% of winners versus 50.2% of losers. The other three traits: owning proprietary data or assets nobody else has (92.9% of winners versus 57.1% of losers), staying tightly focused on a narrow niche rather than a broad category, and carrying a brand name people already recognize.

The traits compound rather than substitute for each other. Sites with none of the five won 13.5% of the time; one trait moved that to 15.4%. Four traits jumped the win rate to 68.1%, and all five to 69.7%. Recognizable brands made up 32.6% of winning sites versus 16.1% of losing sites — roughly twice as common among winners. The unifying idea: Google has stopped rewarding content that explains an answer and started rewarding destinations that let a person finish the job the search implied. That is a harder thing to build than another article, and it is exactly why it still works.

What to do instead: four changes worth making this quarter

The first change is to retire organic traffic as the primary scoreboard for content performance and replace it with a dashboard that correlates brand and visibility signals against revenue: branded search volume, direct traffic, share of voice, assisted conversions. Natividad, who co-authored the research, points to SaaS companies like HubSpot as examples where organic sessions are falling while pipeline keeps growing — a pattern a traffic-only report reads as failure and a revenue-correlated dashboard reads correctly, as a channel mix shifting. The trade-off is real: a correlation dashboard is messier and harder to defend in a board meeting than a rankings report, and it takes longer to prove itself.

The second is to re-segment your content investment by query intent rather than search volume. Branded, local, and high-intent transactional content still converts and deserves disproportionate budget. Informational content should be retooled as citation infrastructure — written to get the fact right when an AI Overview or answer engine pulls from it — rather than judged on sessions it will increasingly not receive. The trade-off: your team will publish fewer broad-appeal explainer posts, and someone will have to explain to a stakeholder why the blog’s session count keeps dropping even as the work continues.

The third is to rebuild your highest-traffic pages so a visitor can finish a task on them, not just read about one. That means calculators, configurators, proprietary benchmarks, or interactive tools where an article currently sits — the same shift Shepard’s data shows separating winners from losers. The trade-off is budget and timeline: this is a product and engineering investment, not an editorial one, and most content teams do not currently have that money in their plan.

The fourth is to invest deliberately in becoming a named, recognizable brand — the trait that, alone, doubles your odds in Shepard’s data — and to convert any attention you earn off your own site into something you own: an email list, a community, first-party data. Building presence on platforms you do not control, including LinkedIn, YouTube, Reddit, and industry newsletters, is necessary now that citations and mentions happen there, but doing it without a plan just swaps a dependency on Google for a dependency on someone else’s algorithm. The trade-off: this work is slow to attribute and hard to report on in a monthly dashboard, which means it needs a sponsor willing to defend it before it shows results.

The writing on the walls for what’s to come

SEO fundamentals are not obsolete. Technical health, site structure, and authority signals still determine whether you rank for the branded and transactional queries that still convert, and they still determine whether an AI system trusts your page enough to cite it. A team that abandons search discipline entirely will lose ground on both fronts.

What has changed is the return on that discipline. It no longer buys you traffic by default; it buys you eligibility to be cited, to rank for the minority of queries that still send clicks, and to be the name a buyer already trusts by the time they search for you directly. Zero-click search will keep climbing — AI Mode alone is growing fast enough that even its current rounding-error share of search will matter within a few quarters — and the marketing teams that fare best will be the ones who stopped measuring the old thing and started building the new one before the board asked why traffic was down.

The practical first step is small: pull the last 12 months of organic sessions, split by branded, informational, and transactional intent, and look at where the decline actually concentrated. If it is broad and even across all three, you have a technical or authority problem worth a full audit. If it is concentrated in informational queries while branded and transactional traffic held steady or grew, you do not have a traffic problem. You have a reporting problem, and the fix is the dashboard, not the content calendar.