A while back I stood in front of a room full of marketers and told them their organic traffic would drop about 20 percent as AI search took hold. I said it with a straight face and a tidy slide.
I was wrong. It’s worse than that, and it’s happening faster than I penciled out.
This isn’t a gradual shift in how people find things. It’s the biggest change to digital marketing since search itself showed up 25 years ago, and the numbers have only bent further the wrong way since I first wrote this down.
The disruption, by the numbers
Start with where the searches are going. ChatGPT alone now handles on the order of 2.5 billion prompts a day. A large share of Americans now treat it like a search engine, and a real slice of them reach for it first, before they ever open Google. The user base went from 100 million to hundreds of millions in about 18 months, faster than any platform in internet history, and it now sits among the most-visited sites on the web. And that’s one tool. Add Gemini, Perplexity, and Copilot and the picture gets steeper.
Now the part that hits your reports: the zero-click search era.
- Over half of Google searches now end without a single click to any website. On mobile it runs higher still.
- When AI Overviews appear, the number one organic result loses roughly a third of its clicks.
- A growing share of consumers now get their answer right there and never leave.
The mechanism is simple, and that’s what makes it brutal. The answer used to live on your page. Now it lives in the box at the top, assembled from your page, and the click that paid your bills never happens. This is the core problem I dig into in my piece on generative engine optimization: the game is no longer ranking for the click, it’s being the source the AI cites.
The casualties are real, and they’re big
This isn’t theory, and it isn’t happening only to small sites. Some of the most quoted names in content marketing took direct hits.
HubSpot, a company that more or less wrote the playbook on inbound, watched its organic traffic fall by about 75 percent over a matter of months, from roughly 13.5 million monthly visits down to about 6 million. If the textbook author can get hit that hard, no content library is safe by default.
The pattern repeats across whole categories:
- News publishers saw organic traffic drop sharply, with the majority of the top 50 domains down year over year, and ChatGPT referrals nowhere near enough to cover the loss.
- Chegg reported steep declines and sued Google over AI Overviews.
- Stack Overflow’s traffic fell by roughly half as developers asked the model instead.
- Even Wikipedia, about as trusted a source as the web has, posted a meaningful decline.
I want to be careful with these figures. Traffic numbers swing, and every one of these stories has nuance. But the direction isn’t in question, and the sites that have reported fresh numbers since haven’t bounced back to where they were.
What I’d do in the next 30 days
Doom isn’t a plan. Here’s the short list I now run with clients before we touch anything fancy.
1. Audit your AI visibility first. Take your top brand and category queries and run them through ChatGPT, Gemini, Perplexity, Claude, and Google AI Overviews. Read what comes back. Plenty of companies find they’re simply absent, or worse, misrepresented. You can’t fix what you haven’t looked at. Tools like AI visibility trackers help, but the manual read is where the gut-check lives.
2. Restructure content so a model can lift it. Models reward content that’s easy to parse and quote: clear, definitive statements, real headers, FAQ blocks, and structured data on your high-intent pages. You’re writing for two readers now, the human and the machine that summarizes you to the human.
3. Fix your measurement before it lies to you. Old metrics like raw CTR and session duration tell you less every quarter, and AI-referred visits often land in analytics as “direct,” so attribution gets murky. Start tracking what actually matters in this channel:
- A custom GA4 segment to flag AI sources, using a referral pattern like
.*chatgpt.com.*|.*perplexity.*|.*claude.ai.*|.*gemini.google.com.* - Leads and sales coming from those new sources
- Whether your brand shows up, and how it’s described, in AI answers
- Your share of voice across ChatGPT, Perplexity, Copilot, Claude, and Gemini
4. Rebalance the budget, on purpose. The old split was something like 30 percent technical, 40 percent content, 20 percent distribution, 10 percent analytics. That math assumed clicks. A starting point I’m testing now, which you should tune to your category:
- 30 percent technical
- 30 percent AI playbook: content built to be cited, structured data, distribution, PR, multi-platform presence
- 30 percent traditional SEO, because it still feeds the AI answers for now
- 10 percent analytics
None of this is a finished answer. It’s a direction, and it beats waiting.
Winners and losers in this round
Here’s the uncomfortable truth I keep landing on. The companies pulling ahead are the ones treating AI search as its own channel, with its own measurement, not as one more line in the SEO report. They’re building brand authority and rebuilding their technical structure even when this quarter’s traffic dips, because they can see where the river is heading.
The ones optimized only for traditional search are already losing share. Most of them just haven’t read it on a dashboard yet.
And the bar isn’t low. Most AI initiatives still struggle to show ROI, which tells you that buying tools isn’t the same as having a strategy. The teams that win this will have a clear plan for the channel, people who actually know how to use the tools, and a habit of testing as the ground keeps shifting.
What’s next
Gartner figured traditional search volume would drop about 25 percent by 2026 on the back of AI chatbots and agents. We’re inside that window now, and from where I sit that estimate is looking conservative, not aggressive.
So three questions worth sitting with:
- When did you last check how the major AI platforms describe your brand?
- What share of your marketing budget is actually pointed at AI visibility?
- If your Google traffic dropped 30 percent next month, would you be ready, or just surprised?
The freight train already pulled into the station. The question was never whether your marketing gets disrupted. It’s whether you adapt before the company across the street does.
If you want the full map of where AI fits across your marketing function, and which parts of search are worth defending first, start with where AI actually fits and then go deep on generative engine optimization.
Get the AI-search reality check each week
I called the size of this shift wrong once, out loud, in front of a room. So now I watch it closely. Twice a week I send what’s actually moving in AI search: the traffic data worth trusting, the moves that are working for real clients, and the ones that flopped so you don’t have to repeat them. If you’d rather see the next disruption coming than read about it after it hits your dashboard, this is the one to be on.