The New Discovery Layer: How AI Is Reshaping Traffic for Publishers

At Nota, we’ve been hearing the same question in nearly every newsroom conversation: “What happens when AI becomes the new front page of the internet?”
That question is no longer hypothetical. Google’s AI Overviews, the rise of AI browsers like Arc and Perplexity, and the growing dominance of platform-based summaries are already reshaping how readers encounter information. For publishers, the challenge is clear: the old discovery pipelines such as search, social, and direct are fragmenting under the weight of a new, AI-driven layer.
The Shift in Discovery Patterns
For two decades, publishers oriented their content strategies around SEO and social virality. But AI is now inserting itself between audience and publisher. Readers are asking AI systems for “the best local news on housing policy” or “summarize the Yankees’ last five games,” and increasingly, the AI response is the destination itself.
The numbers tell the story. Between May and June 2025, members of Digital Content Next — including The New York Times, Condé Nast, and Vox — reported Google AI Overviews causing 1% to 25% declines in search referral traffic, with a median year-over-year drop of around 10% (–7% for news, –14% for non-news). Ahrefs data shows that when AI Overviews appear, the first organic link loses an average of 34.5% of clicks, and some publishers, like Business Insider, have seen losses of up to 55%.
Meanwhile, AI browsers and aggregators are experimenting with attribution differently. Perplexity, for instance, has begun sending a small but noticeable trickle of traffic — though across generative platforms overall, only 3%–7% of referral links go to news content. The discovery funnel is no longer linear. It is being mediated by opaque algorithms designed to summarize first and send clicks second.
Risks and Opportunities
The risks are obvious:
- Reduced referral traffic — Some outlets are already seeing double-digit declines, with click-through rates dropping by over a third when AI Overviews appear.
- Brand dilution — Publishers’ work is often summarized without clear attribution, leaving readers with the information but not the source.
- Unpredictable discovery — AI systems surface articles based on training data and prompt design rather than traditional ranking factors.
But there are also opportunities:
- Packaging for AI — Publishers who structure content with clarity, metadata, and context may be more visible to AI systems.
- Trust as a differentiator — As misinformation proliferates, recognizable brands carry weight inside AI responses.
- New distribution partners — Some AI browsers are testing explicit attribution, licensing, and even revenue-sharing models.
What We’re Hearing From Newsrooms
Traffic forecasts are harder to model. Attribution policies change week to week. Yet, there’s cautious optimism. At Nota, our editorial team frames it this way: the industry has survived social pivots and algorithm updates before — and it will adapt again. The difference this time is urgency: publishers need a playbook, fast.
How Nota Thinks About It
We see our role as helping publishers adapt responsibly to this shift. That means:
- Assistive AI tools that package stories into clear, context-rich formats AI systems can understand.
- C2PA compliance so content carries transparent signals of trust and authorship into AI environments.
- Workflow efficiency so stretched teams can focus on reporting while AI helps with packaging and repurposing.
The discovery layer is changing, but the core mission of journalism has not. Informing, engaging, and earning audience trust remain central. AI may reshape the front door, but publishers who adapt early can still own the living room.
Looking Ahead
The next year will be decisive. AI systems are racing ahead, regulators are circling, and readers are forming new habits. For publishers, the choice is whether to be shaped by this discovery shift or to shape it themselves.
At Nota, we are committed to making sure newsrooms do not just keep up, but lead.