The single number I'd put on the wall of every marketing meeting in 2026: 68% of Google searches now end without a click.
That's up from 45% ten years ago. For every 1,000 US Google searches, only 320 clicks go anywhere at all. And of those 320 clicks, only about two-thirds actually leave Google's own properties.
If you've been dependent on organic search traffic for the last decade, this is the underlying number that decides whether your marketing strategy still makes sense. And AI didn't cause it. AI is finishing what a ten-year trend was already going to do.
Here's the strategy I'd run if I were rebuilding a marketing motion from scratch today. Seven sections on what actually changed, plus where I'd start.
The math that broke website traffic
Before we talk about what to do, let me lay out the numbers so the "why" is unambiguous.
Zero-click rate on Google search over time:
- 2016: 45%
- 2020: about 50%
- 2023: about 58%
- 2026: 68%
For every 1,000 US Google searches today:
- 390 end without any action
- 290 end with the user typing another search
- 320 end with a click
Of those 320 clicks:
- Around 60 go to a paid ad
- Around 85 go to AI mode or another Google-owned property
- Around 210 go to the actual open web
Meaning the assumption a decade of SEO strategy was built on (that a Google search meant a click to your site) is now roughly 21% true. Four out of five Google searches don't produce any open-web traffic at all.
The second number worth carrying: only about 3% of searches happen inside AI tools like ChatGPT or Perplexity. Google is still where 80% of search behavior lives. E-commerce sites and social media catch most of the rest. The most-cited version of the "AI is eating search" narrative overweights AI-tool share and underweights the fact that Google's own zero-click behavior is doing most of the damage.
Two implications for your strategy. First, if you've been waiting for organic traffic to recover as AI settles down, it isn't going to. The mechanism destroying your traffic is the growing pile of features Google now serves inside the search results page itself: featured snippets, direct answers, knowledge panels, AI mode, ads. Second, even in a world where AI eventually gets more search share, Google is still your largest single referrer for the foreseeable future. Ignoring it is the wrong move. Depending on it the way you did five years ago is also the wrong move.
Zero-click didn't start with AI. It just finished with it.
The story people tell about the last three years: ChatGPT launched, AI ate search, marketing broke. Neat, and mostly wrong.
Zero-click has been growing since 2016. Google spent the decade before ChatGPT launched systematically keeping users on the search results page. Featured snippets in 2016. Local packs, weather cards, direct answers, rich cards, knowledge panels through 2018-2022. Every one of them was a chunk of your traffic that Google kept for itself.
What AI did: it accelerated an existing trend and gave people a story to make sense of it. The trend didn't need AI. By the time the zero-click rate ticked past 60%, most sites' organic-traffic economics were already broken. The AI overlay just made the problem visible.
Two implications from getting the causality right.
There's no version of this where the rate goes back below 50%. The direction has been one way for a decade, and every product improvement Google ships makes zero-click larger, not smaller. Waiting for the wave to recede is waiting for a wave that isn't leaving.
Abandoning search entirely and going all-in on answer-engine optimization also doesn't work. AI tools account for 3% of searches. Optimizing entirely for 3% of a shrinking pie is a worse call than optimizing for the fragmented but larger reality of Google plus TikTok plus Amazon plus Reddit plus YouTube plus a few review sites plus AI tools.
The right frame is that search is diversifying. Your marketing motion has to work across the fragmentation, not pick one side and hope.
HubSpot lost 140 million visits and revenue hit an all-time high
This is the story I'd force every marketer to internalize.
HubSpot lost roughly 140 million visits over a 12-month period. Around 80% of organic traffic gone. The SEO community treated it like a corporate collapse in slow motion. Then Q4 2025 landed and HubSpot reported all-time-high revenue.
What actually happened: the pages that lost traffic were the ones you'd expect if you're honest about the old inbound playbook. Movie-quote listicles. Emoji lookup articles. Everything that ranked because HubSpot's domain authority let it rank for anything marginally related to how someone might one day become a customer. All of that vanished. The pages that mattered (CRM, marketing automation, category-defining software terms) held on.
The old inbound playbook was volume-driven. You wrote for a thousand keywords adjacent to your ICP, converted a small percentage of them, and the pipeline compounded. It worked because Google was happy to serve a wide range of queries out to the open web. Conversion rate on low-intent traffic was maybe 0.5%, but when your monthly traffic was bigger than TechCrunch, half a percent was a real business.
That model is over. The 0.5%-on-low-intent-traffic channel is closed for everyone. What's left is a much smaller pool of high-intent search demand and a much larger pool of channels-that-aren't-search that you have to build audiences on.
The SEO that survives is small, sharp, and category-focused. You want to rank for the 30-50 terms a serious buyer types when they're actively evaluating. Everything else in your marketing surface has to build the right audience in the places they already spend time.
The revenue-up-while-traffic-down story only makes sense if you understand which traffic you lost. HubSpot didn't lose the CRM buyer. It lost the shrug-emoji searcher. Everyone else running the same volume-first playbook is losing both, because they never built a moat on the terms that matter and their low-intent volume was propping up the whole funnel.
Building an audience vs building the right audience
There's a difference between having a big audience and having the audience that actually buys from you, and it's the biggest strategic split in marketing right now.
Big audiences are cheap now. You can spin up a TikTok or LinkedIn account, chase whatever the algorithm rewards this quarter, and pull hundreds of thousands of followers in six months. The problem is that an audience that came for viral takes on any topic doesn't buy your specific software.
The horizontal-creator play (Diary of a CEO is the canonical example) works when the goal is ad revenue on watch time. If you're selling a product, virality-driven horizontal audiences convert at approximately zero. You spend months building reach and can't convert any of it because the audience's interest didn't match your offer.
The version that works is going deep in a niche. Not "marketing." Not "business." Something like "GTM for infrastructure companies in the seed-to-Series-A range" or "SEO for e-commerce brands with under $5M in revenue."
Two reasons depth wins right now:
- AI covers breadth for free. Nobody needs a horizontal generalist creator to summarize the news of the week when Claude or ChatGPT can do it in three seconds. What people still need is someone who has thought harder about one specific problem than everyone else.
- The follower-to-customer conversion is dramatically higher on a niche audience because they came to you for exactly the problem you solve.
The tactical move that clicks: start extremely deep. Once you own the niche, you can broaden. Codie Sanchez is the canonical example. She spent years known specifically for buying boring businesses (laundromats, car washes), then expanded into business ownership generally, and now sits on a much wider audience while still holding the depth reputation. Starting with generalism and trying to earn depth later doesn't produce either.
For a founder reading this: the honest question isn't "which channel should we be on." The honest question is "what is the single specific slice of a category we want to own, and where do the buyers in that slice spend their time." Everything else falls out of that answer.
Where your people actually are: channel → source → tactic
"Go where your audience is" is hollow advice unless it comes with a method. Here's the ladder I run.
Channel. Which platforms reach your specific ICP? Not "LinkedIn is a B2B channel." Which specific channels does the CTO of a fintech Series B actually open before 10am. This is a research question, and there are tools (SparkToro is the one I reach for) that pull search plus clickstream plus social data to tell you which platforms your defined audience over-indexes on relative to the average web user.
Source. Once you know the channel, figure out the specific sources on that channel that shape the audience's thinking. Which creators, publications, podcasts, subreddits, Discord servers, and newsletters do they consume. The channel tells you where to be present. The sources tell you who you need to be in the room with.
Tactic. Once you know the sources, the tactics fall out. Co-marketing partnerships with the top three creators. Sponsored posts in the top two newsletters. Regular guesting on the top podcast. Digital PR into the two publications your buyer actually reads.
Two filters most people skip.
Team strength. Whatever platform or format the ladder points at, look at whether your team can actually execute on it. If you have strong writers but nobody who's comfortable on camera, forcing a YouTube strategy will produce mediocre content nobody watches. If you have a founder who's brilliant on podcasts but freezes at written long-form, that founder should be running a podcast guest tour, not producing ghostwritten LinkedIn posts. Play to what the team can do at world-class quality.
Enjoyment. If you or your team hate being on a platform, you won't be good at it. This is the piece nobody puts in strategy docs because it sounds unserious. It's the piece that decides whether a channel actually performs a year later. Every content operation I've watched slowly die was one where the team resented showing up.
The ladder gets you 80% of the strategic answer. The two filters keep the answer from being one you can't actually execute against.
Zero-click value: give it native to the platform
The last tactical shift worth naming: in a zero-click world, your content has to deliver value that stands on its own inside the platform. The old model of teasing content and hoping viewers click through to your site for the payoff is finished. Only about a third of viewers ever click through anything anymore.
The canonical example is Duolingo's TikTok. You can watch all of their content, laugh, become a fan of the brand, and never once download the app. The zero-click value is entertainment, and the entertainment is complete inside the platform. Downloads happen anyway because the audience trust compounds over months of enjoying the content.
For a B2B example: a founder who writes threads on LinkedIn that fully answer the question in the thread itself, without the "book a call to hear the rest" cliffhanger. The value is delivered where the audience is. The commercial outcome (booking calls, buying software) happens as a downstream consequence of trust.
This is where "value debt" enters. When people consume value from you for months without paying anything, a psychological weight builds. At some point they feel they should support you, buy something, or refer someone who will. Hard to measure. Real in effect. Most of the commerce I've watched work in the last year runs on some version of this dynamic.
The mistake is trying to shortcut the value debt. Publish thin content, put the real substance behind a signup wall, expect the conversion. That worked in 2016. In 2026 it gets ignored because the audience has infinite better options one platform over. Publish the substance. Publish it native to the platform. Let the audience build up value debt over months. The conversion comes back cleaner because the audience trusts you specifically instead of clicking through a funnel.
What I'd rebuild first
If I inherited a marketing motion tomorrow and had a quarter to reset it for the zero-click era, the order I'd run in:
Audit the traffic you have and split it by intent. What percentage of traffic is high-intent to your ICP? What percentage is HubSpot-style low-intent volume that used to convert at 0.5% but doesn't anymore? Kill or deprioritize the second bucket honestly.
Pick the specific slice you want to own. Write the ICP down in one sentence. If it's more than one sentence, keep narrowing. A slice you can own is worth more than a market you can't.
Run channel → source → tactic on that slice. Figure out the actual channels, the actual sources, and the actual tactics. Layer the team-strength and enjoyment filters on top.
Publish native value on the top one or two channels. Concentrate on one channel at world-class quality rather than five channels at half strength. Six months minimum before you evaluate.
Instrument measurement, not attribution. You can't attribute value debt. You can measure whether pipeline moves. Track leading indicators (audience growth on your one channel, share of voice in your niche, inbound demo requests that name your content) rather than click paths and last-touch conversions.
The whole thing is essentially picking a smaller game to win at cleanly, in a market where the volume plays no longer work. Which is the shape of every good strategic move in a saturated market.
If you want to see the wider stack I run around this pattern, the current AI stack lives here. More posts in this shape live in Writing.