Search doesn't work the way it did five years ago, and the numbers make it hard to argue otherwise. More than two-thirds of Google searches now end without a single click. AI tools like ChatGPT and Claude keep fewer than one in a thousand visitors on their own results. Social platforms hold onto 98 or 99 out of every 100 people who start a session there. Whatever we thought "search" meant, it has quietly become something else: an activity that happens everywhere, on every platform, most of the time without ever sending anyone to a website.
We recently talked with Rand Fishkin, co-founder of the audience research tool SparkToro, about what this shift actually means for marketers. His take is blunt: the old playbook isn't just declining, it's over. "It's like we're in 2010 and you are optimizing for Hotbot and Lycos," he said. "The time has passed. We need to move on and recognize that the instant answer, the zero-click era, is formally here."
The Numbers Behind the Shift
According to research Fishkin's team completed with similarweb, 58% of Google searches ended without a click in 2024. That number is now above 67%, the fastest two-year jump in the twelve years his team has been tracking it. The pattern holds well beyond Google:
- AI tools send fewer than 1 in 1,000 visits back out to the open web.
- Social platforms (Instagram, LinkedIn, Reddit, TikTok, YouTube, Facebook) keep 98 to 99 out of every 100 users inside the app.
- Instant answers now show up directly inside searches on Instagram and TikTok, not just Google.
The term for this, coined by a UK-based marketer named Ashley, is search everywhere optimization. As Fishkin put it, "search is way bigger" than Google, Bing, or DuckDuckGo. It happens on YouTube, on Reddit, inside every social app and every e-commerce site. Treating SEO as a Google-only discipline is already behind the curve.
Attribution Was Always an Illusion
Here's the harder pill to swallow: even before the zero-click era, precise attribution was mostly a story we told ourselves. Fishkin compared today's marketing landscape to the ad agencies of the 1960s, who never pretended to trace a single sale back to a single billboard. They measured lift in same-store sales across a region instead. "No one even tried" individual-level attribution, he said, because it was obviously unrealistic at scale.
For a strange stretch of about fifteen years, roughly 2005 to 2020, that changed. Dashboards could show a clean path: a LinkedIn visit led to content consumption, which led to an email signup, which led to a conversion. It felt like real measurement. But that window is closing fast. LinkedIn now hides referral data on a large share of organic visits. People increasingly consume content without ever leaving the platform. Posting a link limits your reach; skipping the link protects your reach but erases your attribution trail entirely.
The result, in Fishkin's words, is that "we still haven't evolved past" the belief that everything can and should be attributed. That belief has quietly pushed budgets in one direction.
Paid Gets the Credit. Organic Does the Work.
When you buy an ad on LinkedIn, Google, Meta, Reddit, or TikTok, you get a click, and with it, clean attribution that looks great in a dashboard. Organic influence, meanwhile, often shows up nowhere. Fishkin argues this asymmetry, not actual performance, is why paid budgets have exploded: "Organic might be ten times, a hundred times the influence. Paid is always going to get the credit."
He estimates the real split in influence is still roughly 80/20 in favor of organic. But because paid is measurable and organic often isn't, the incentive structure inside most companies rewards the channel that looks good in a report over the channel that's actually doing the work.
Big platforms have little reason to correct this. As Fishkin noted, Google, Meta, and other major players benefit when marketers believe attribution still functions cleanly, because that belief keeps budgets flowing toward channels that show up as "converting traffic," which tends to mean paid.
Short-Form Video Is the New Superpower
If precise attribution is gone, what should marketers actually invest their energy in? Fishkin's answer is direct: short-form video. "Nothing else captures attention like short form video," he said, "and this is regardless of platform." YouTube Shorts, TikTok, Instagram, a well-placed video on the right subreddit, even short video on LinkedIn all perform at a level nothing else matches right now.
He frames this as a meaningfully more creative discipline than traditional SEO ever was. Optimizing a meta description or chasing a featured snippet was "mildly creative but pretty technical." Short-form video demands a much deeper read of human psychology, attention, and hooks, with an iteration cycle far faster than almost anything that came before it. Multiple videos can go out in a single day, each one a fresh test of what actually holds attention.
Distribution Beats Optimization: What Reddit Taught Us
One of the more counterintuitive lessons from our conversation was about Reddit. Fishkin has started posting his own research charts directly into relevant subreddits, and the results surprised even him. A piece of zero-click search research that reached roughly 100,000 people on LinkedIn and a few thousand on the company blog pulled in more than a million impressions on Reddit.
Almost nobody clicks through. "It's AI levels of click-through rate," he said, "like one in a thousand people." But that's not really the point. Reddit doesn't drive direct signups. It drives visibility among journalists and other people who shape opinion, and that secondhand coverage lifts brand search volume, AI visibility, and general awareness in ways a direct click never could.
The framework underneath this: every business has two audiences, the customers who will eventually buy, and the influencers who shape what those customers believe. Reddit, for Fishkin, is squarely an influencer channel. Platforms like LinkedIn, threads, and email newsletters do the work of converting the people already close to a purchase decision.
Brand Recognition Is the Long-Term Play
The clearest example of where all this points came up almost as an aside. Looking at a competitive analysis in the active wearables space, one brand stood out for doing none of the modern AI visibility best practices and still dominating: Garmin. Its advantage isn't technical optimization. It's simply the best product and the strongest brand, full stop.
Fishkin sees this as the throughline for everything else in this conversation. In a world where individual clicks and precise attribution are disappearing, brand recognition becomes the asset that actually compounds. It's also the logic behind tools built to track brand mentions across the web, on the theory that AI models are trained on exactly this kind of co-occurrence data, brand plus topic, showing up again and again across sources. Short-term tricks may still work for a few months, the same way spammy backlinks once worked for Google. But the channels built on real recognition, video, LinkedIn, Reddit, YouTube, are the ones likely to keep paying off five and ten years out.
The hardest part isn't understanding any of this. It's convincing the people who control budget to act on it. As Fishkin put it, it's oddly easy to get executives to throw money at the newest AI visibility trend, and strangely difficult to get the same budget approved for a YouTube strategy or a long-term content program, even when the second option is very likely to matter more.
