We've read the reports. We've read the stories. We've pulled out the nonsense. What follows is everything you need to know about what's actually happening to publisher traffic right now. The facts. Nothing more.

Dave Bouskill and Debra Corbeil spent 16 years building The Planet D into one of the most recognized travel blogs on the internet. Then Gemini rolled out. Traffic dropped 90% in six months. They laid off staff. They shut down the website.

This is not an outlier. This is the new normal.

Google Search referrals to news publishers collapsed from 51% to 27% in two years (Reuters Institute, 280 executives, 51 countries). Those same executives expect an additional 43% decline over the next three years.

People Inc., formerly Dotdash Meredith, lost 50% of its Google search traffic in two years. Google went from 70% of their traffic to about 30% (IAC Q4 2025 earnings call).

US publishers saw a 38% year-over-year decline in Google Search traffic (Chartbeat, 2,500+ websites). Globally the number was 33%.

Google Discover collapsed 21% globally and 29% in the US year-over-year. One publisher managing four news websites went from over 100,000 daily Discover clicks to zero overnight after the December 2025 core update (Reuters Institute).

UK website traffic growth collapsed 86% since Google's AI search rollout (Tank research, 800 companies, 16 sectors).

AI Overviews reduce organic clicks by 34.5% to 54.6% when present in search results (Ahrefs, 300,000 keywords). Only 1% of users click links when AI Overviews appear (Pew Research).

Bot traffic now exceeds human traffic at 51%+ of all web traffic (Infosecurity Magazine). AI crawl-to-visit ratios have hit 250:1 and OpenAI's reached 1,250:1 (Cloudflare). They're taking everything and sending back almost nothing.

The Spectator lost 64% visibility during the December core update. The Telegraph fell 30%. Reuters fell 31% (SISTRIX). India's Hindustan Times dropped from above 6 points to below 2 in 18 days.

This is not cherry-picked. This is the entire industry.

Here is how the web used to work: Search engine crawls your content. User searches. Search engine returns links. User clicks. User lands on your site. You show ads. You get paid.

Here is how it works now: AI engine crawls your content. User asks a question. AI synthesizes an answer from your content and a dozen other sources. User never visits your site. AI platform shows ads around that answer. You get nothing.

Your content is the fuel. The AI companies are the engine. They're not paying for gas.

IAB CEO David Cohen at the 2026 Annual Leadership Meeting: "Free riding isn't just unfair. It's stealing. Unless you pay for content that AI bots scrape, you will ruin the economic model that makes the content available in the first place."

Google Discover now accounts for 67.5% of publisher traffic from Google (Reuters Institute). But Discover is volatile, unpredictable, and entirely passive. You don't rank in Discover. You get chosen by an algorithm. One day you get 100,000 clicks. The next day, zero. That's not a business model. That's a slot machine.

YouTube is now the #1 platform priority for publishers at +74 net score. AI platforms are #2 at +61 despite delivering less than 0.03% of actual traffic (Chartbeat). Traditional Google SEO scored negative 25 on the publisher priority index (Reuters Institute).

The 20-year era of optimizing for Google blue links is over.

Less traffic means less revenue. Less revenue means less content. Less content means less raw material for LLMs. Worse training data means worse AI outputs. Worse outputs mean a degraded information ecosystem for everyone.

France has documented over 4,000 fake news sites powered by generative AI (NewsGuard). Zombie local news sites in the US now outnumber real local news outlets. Nearly 1 in 10 of the fastest-growing YouTube channels show only AI-generated content (Reuters Institute).

Even the AI companies need this to get fixed. They are eating the seed corn.

Executive confidence in journalism's future dropped from 60% to 38% in four years (Reuters Institute). Publishers are abandoning the content categories AI can replicate: service journalism scored -42, general news -38, evergreen content -32. They're retreating to what AI can't fake: original investigations +91, contextual analysis +82, community building +76 (Reuters Institute).

76% of publishers now plan to encourage journalists to behave more like creators. Video is the future for 79% of executives. Only 20% said text would become more important (Reuters Institute).

The industry knows where this is going. The question is whether it moves fast enough.

People Inc. lost half its search traffic. And grew digital revenue 14% year-over-year to $355 million (IAC Q4 2025 earnings). Off-platform views nearly doubled in two years. Non-session revenue hit $137 million in Q4, up 37%.

Curiosity Stream reported 425% year-over-year growth in LLM licensing revenue (industry analyst reporting). The IAB is building a protocol called CoMP that could create a standardized "door with a lock" for publisher content. LLM operators are already working with rate cards in the $10 to $30 CPM range at scale.

The money is there. The demand is real. Most publishers just don't have the plumbing to capture it yet.

This is the free preview. The blueprint is one thing. The operating system underneath it is another.

In Part 2, we break down exactly who's getting paid, how much, and through what mechanisms. The specific deals. The actual revenue lines. The models that are already working while most of the industry is still arguing about robots.txt.

In Part 3, we give you the playbook. The five moves every publisher needs to make right now. CoMP, rights infrastructure, content marketplaces, and how to turn your "contribution" into a compensated revenue line before the window closes.

The old web economy is over. The new one is being built right now. The only question is whether you're building it or being consumed by it.