Start by accepting the premise the data forces on you: the clicks AI Overviews removed are not returning, because the answer is now delivered on the results page instead of on your site. That changes the job from winning the click back to three things you can still influence. Defend the visits that remain high in intent. Become the source AI systems cite when they answer in your place. And earn from the AI agents that now read your content directly. Running the old SEO playbook harder will not reverse a structural change in how discovery works, but treating it as a different game can protect, and partly replace, what was lost.
How much traffic has actually been lost
The decline is large and well documented. Analyses through 2025 and into 2026 found Google search traffic to publishers down by roughly a third globally, with steeper falls in the United States than in Europe. When an AI Overview appears, the zero-click rate, the share of searches that end without any click to a website, rises sharply, and click-through to the links cited inside an Overview is low. News sites have reported referral declines in the range of a quarter or more in the year after AI Overviews rolled out, and individual large publishers have reported drops of half or worse on some content. Crucially, referrals from AI chatbots have not filled the gap; combined, they account for a very small share of publisher pageviews, far below what Google search used to send.
Why the old traffic is not coming back
AI Overviews and AI assistants change the deal. Search used to be a directory that sent users onward to publisher pages, where the publisher monetised the visit. Increasingly, the answer is assembled and delivered in the search or chat interface itself, drawing on publisher content without sending the user to read it. Informational queries, which historically delivered the best display CPMs, are the easiest for an AI to answer in place, so they are the first to lose their clicks. Optimising harder for a position that users no longer click does not restore the revenue.
The three-part response
Protect the high-intent traffic that remains
Not all traffic is equally exposed. Transactional and high-intent pages, where a user wants to compare, buy, or act, still attract clicks because the user needs to complete something a summary cannot finish. Concentrate effort there, strengthen those pages, and accept that thin informational content is the most cannibalised category. Build direct relationships through newsletters, apps, and registered audiences so that some of your reach no longer depends on a search referral at all.
Become the source AI systems cite
If users get their answer from an AI, the strategic goal shifts from ranking for clicks to being the source the AI cites. This is the discipline known as generative engine optimisation. Content that gets cited tends to be well structured, factual, clearly attributed, and recently updated, with direct answers near the top and a clean parseable structure. Being cited keeps your brand in front of the user even when the click does not happen, and it is the foundation for the third part of the response.
Open a revenue line from AI agent traffic
The same shift that took your search clicks also sends AI agents to read your content directly. Live Search Agents from ChatGPT, Perplexity, Claude, and Gemini fetch your pages in real time to answer user questions, and today that read generates nothing because the agent runs no JavaScript and triggers no ad. This is a new, growing audience arriving at the moment of purchase research. It can be monetised. blankspace intercepts Live Search Agent retrievals at the CDN edge and runs a real-time auction to inject a relevant brand fact into the content the agent reads, attributing revenue the same day. It does not recover the lost human clicks, but it turns the AI traffic that replaced them into income rather than cost.
What to measure now
Standard analytics understate the problem because they cannot see agents that do not run JavaScript. Add server-side measurement so you can see AI agent traffic alongside human traffic. Track which pages AI systems retrieve most, whether your content is being cited in AI answers, and how your high-intent pages are holding up versus your informational pages. These metrics tell you where to defend, where to optimise for citation, and where the new revenue opportunity is largest.
Frequently asked questions
Is the traffic lost to AI Overviews recoverable?
Not in its previous form. AI Overviews answer queries in place, so the clicks they remove do not return through better SEO. The recoverable value is in defending high-intent pages, getting cited in AI answers, and monetising AI agent traffic directly.
Do AI chatbot referrals make up for lost Google traffic?
No. Combined referrals from AI chatbots are a very small fraction of publisher pageviews, far below the volume Google search previously sent. They are growing but do not offset the decline.
Which content is hit hardest by AI Overviews?
General informational and how-to content is hit hardest, because an AI can answer it in place. Transactional and high-intent pages retain more clicks because users still need to complete an action.
What is the difference between AI Overviews traffic and AI agent traffic?
AI Overviews reduce clicks to your site from Google's results page. AI agent traffic is the separate stream of bots that fetch your pages directly to answer user questions in tools like ChatGPT and Perplexity. The first is a loss to manage; the second is an audience to monetise.
How do I monetise the AI agents reading my content?
Through content-layer advertising at the CDN edge, which inserts a paid contextual brand mention into the text an agent retrieves. Because agents skip JavaScript, this is the layer that can earn from them, and it requires no change to your CMS or human-facing ad stack.