Have you heard your SEO agency say, “Generative AI doesn’t affect your SEO”? Maybe they told you not to worry—but let me be blunt: they’re lying—and here’s why you’re about to understand in depth. As ChatGPT referrals to news outlets climb, organic search traffic is sinking. Let’s examine this seismic shift, break it down clearly, and help publishers like you adapt smartly.
ChatGPT, Generative AI, and Publisher Traffic — The Changing Landscape
From Search to Answers — Google’s AI Overviews Diminish Click-Throughs
Since May 2024, Google’s AI Overviews have been summarizing news content directly on the SERP, increasing “no-click searches” dramatically—from 56% up to ~69% as of May 2025 (Source: Similarweb). That’s nearly 7 out of 10 news queries ending without a single click to your own site! Naturally, this has slashed organic traffic: visits dropped from 2.3 billion at peak mid‑2024 to under 1.7 billion by May 2025.
Bottom line? If your SEO plan still hinges on bulk traffic from Google keywords, you’re watching your business bleed.
ChatGPT Referrals Rise—But Still Trail Behind Search Declines
Enter ChatGPT. Publishers hoping AI chat tools would save them are seeing some light at the end of the tunnel. News-related ChatGPT prompts grew by a staggering 212% between Jan 2024 and May 2025.
Referral numbers? They skyrocketed from just under 1 million in early 2024 to over 25 million in 2025—a 25× increase. That sounds great—until you compare it to search losses. A drop of ~600 million visits vs. a gain of ~24 million? Not enough.
“ChatGPT referrals grew 25× from early 2024 to 2025, yet offset only a fraction of lost search traffic—AI adoption isn’t traffic salvation, it’s a wake-up call.”
Why Agencies Minimizing Generative AI’s Impact Are Misleading You
Let me put it simply: If your agency says AI doesn’t hurt your SEO, they’re sugarcoating it. The proof is everywhere:
- Search traffic decline is real—Google’s AI Overviews are rapidly reducing click-throughs from the biggest traffic source.
- ChatGPT referrals, while impressive, still only scratch the surface. They offset maybe 5–10% of lost search traffic—not nearly enough.
- Many publishers (like NYT) even see lower AI referrals, possibly due to legal issues or cautious licensing.
Agencies relying on stale SEO tactics are failing you. The match has changed—like insisting on oil lamps in the age of LEDs.
How ChatGPT Referrals Work — Tech & Pricing Breakdown
The Mechanics Behind ChatGPT Referrals
- A user prompts ChatGPT for a news-relevant query.
- ChatGPT (with web browsing or Edge plugin enabled) sources external links.
- It crafts a summary, embedding citations—e.g., “According to Reuters…,” then links to publisher pages.
- If your site is linked, that click counts as a ChatGPT referral.
This system is great for content discovery, but…
Costs Publishers Face in This AI Ecosystem
- Tech integrations: Many must pay for data-tracking tools like Similarweb AI-referral analytics (enterprise costs).
- Content strategy tuning: Requires structured metadata, linkable leads, and strong journalistic hooks.
- Potential licensing fees: If ChatGPT indexes premium content, licensing deals may be needed.
And don’t forget: ChatGPT referrals still trail behind search losses. You invest resource‑intensive efforts for a fraction of the return. But that’s just the baseline; more on fixing that soon!
Similarweb Insights — Who’s Winning, Who’s Losing
Similarweb data paints an enlightening picture:
- Winners: Sites like Reuters, NY Post, Business Insider saw ChatGPT referral increases of 8.9%, 7.1%, and 6.5% respectively.
- Losers: The New York Times, due to copyright issues, only saw a 3.1% uptick.
This divergence underscores that your AI traffic performance depends on positioning, openness, and site structure.
AEO & GEO – Your Blueprint for AI-Driven Visibility
What AEO/GEO Actually Mean for Your Content Strategy
- AEO: Concise, accurate Q&A-style content for AI tools (e.g., “Quick Tips”).
-
GEO: Optimizing metadata, schema,
llms.txt
, and citation formats. - Metadata: Strong headings, bullet lists, and author credentials.
- Citations: Be a trustworthy, frequently linked source.
Implementing AEO/GEO — Practical Tips
Action | Implementation |
---|---|
Add structured Q&A sections | Use H2: “What is generative AI?” + clear answer |
Use schema for articles | Include author, datePublished, publisher |
LLMS.txt file in root | Signal to AI aggregators “I want to be cited” |
Optimize headlines & intros | Make them summary-worthy |
Monitor AI referrals | Use Similarweb or GA with AI source tags |
Real-World Examples — Movers & Shakers
Reuters, NY Post, Business Insider – The Early Movers
- Open-format content on trending topics.
- SEO teams implementing AEO strategies.
- Less cautious about licensing, allowing easier AI access.
What Publishers Should Do Now
🔍 1. Audit & Adapt Content Structure
- Break content into AI-friendly chunks—facts, Q&A, TL;DR bullets.
- Update schema and metadata.
🤝 2. Build Strategic Partnerships
- Explore licensing with OpenAI, Microsoft, Google Gemini.
- Offer structured APIs or RSS for AI ingestion.
📈 3. Track & Optimize Referral Performance
- Use analytics to monitor ChatGPT referrals.
- A/B test with vs. without
llms.txt
.
💡 4. Diversify Audience Channels
- Invest in newsletters, podcasts, social micro-content (TikTok, Instagram).
🧠 5. Lean into Brand Authority & Trust
As AI can hallucinate, credibility and citation consistency matter. Be the reliable source.
Final Takeaway — The New SEO Era Has Arrived
On vous dit que l’IA générative n’a aucun impact sur le SEO? Peut-être votre agence SEO vous l’a dit ? On vous ment ! It does matter. The data is crystal clear:
- Organic search referrals have plunged—thanks largely to Google’s AI Overviews.
- ChatGPT referrals are growing fast—but only replace ~5–10% of lost traffic.
- To compete, embrace AEO/GEO, measure results, diversify your traffic mix.
Useful tips: audit schema, use llms.txt
, monitor referrals, diversify channels, and strengthen your brand signal.
Let’s Keep the Conversation Going!
How are you adapting your content strategy in the AI age? Are you tracking AI referrals yet? Share your challenges and wins below—let’s learn from each other. 💬