The AI death spiral: what it means for your brand visibility

The Bloomberg headline is blunt: "Why an AI death spiral threatens the internet." The argument is straightforward. Publishers create content. Search engines index it, send traffic, and advertisers pay for eyeballs. That loop has funded online journalism and content creation for two decades.

AI-powered search is breaking the loop.

When ChatGPT, Perplexity, or Google's AI Mode answers a query directly, users don't click through to the source. The publisher gets no traffic, no ad revenue, and no reason to keep creating the content the AI model trained on in the first place. Rand Fishkin has called this a "zero-click" crisis. Rutgers professor Caitlin Petre calls it a threat to the long-term economics of journalism.

But this is not a story about death. It's a story about bifurcation. Some brands are adapting. Most aren't.

The case study worth dissecting is People Inc. (the media brand, not a composite). Its CEO Neil Vogel has publicly stated that his company has offset declining search traffic through direct audience relationships and brand licensing. That's the playbook. Let's break it down.

The problem: People magazine and the traffic cliff

People Inc. built its digital business on the same foundation as every major publisher: high-volume Google traffic, display advertising, and content volume. At peak, that model worked. Google sent millions of readers to celebrity news, human interest stories, and lifestyle content every month.

Then two things happened simultaneously. Google's AI Overviews began answering entertainment queries inline. ChatGPT started summarizing celebrity news without attribution. Perplexity began pulling People's own reporting and serving it without a click.

According to Datos and SparkToro research cited by Rand Fishkin, approximately 58.5% of Google searches in the US now end without a click. For informational queries, the type that dominate entertainment media, the number is considerably higher.

For People, this wasn't a gradual decline. It was a structural disconnection between content production cost and content distribution return.

What they changed: direct audience, brand licensing, and AI-native formats

Neil Vogel's response, as reported by Bloomberg, had three concrete components.

First, People doubled down on owned audience channels. Email newsletters, push notifications, and app-based consumption all bypass AI-intermediated search entirely. When a reader has a direct subscription relationship with People, no AI engine sits between the publisher and the revenue.

Second, People accelerated brand licensing. The People name carries 50 years of trust equity. That equity translates into licensing deals, branded events, and commerce partnerships that generate revenue entirely outside the traffic-advertising loop.

Third, and most relevant to the GEO conversation: People restructured content formats to function inside AI citation flows rather than fighting them. Listicles became structured data. "Best of" articles got explicit schema markup. Staff bylines were reinforced with author E-E-A-T signals across multiple external platforms. The goal shifted from "rank for this keyword" to "get cited as the authoritative source when an AI answers this question."

This last move is the one most brands haven't made yet.

The results: before/after

Vogel hasn't released hard traffic numbers publicly, but the directional signal is clear enough to analyze structurally.

Publishers who have pivoted to direct audience and AI citation strategies are seeing newsletter open rates climb as search referral traffic falls. BrightEdge's 2024 channel shift data showed that organic search's share of trackable traffic dropped from 53% to 44% between 2020 and 2024, while direct and email traffic held steady or grew. For publishers making the pivot early, the revenue per reader from owned channels is materially higher than from search-referred advertising.

More relevant for GEO practitioners: structured, citable content from authoritative publishers gets cited by AI engines at significantly higher rates than unstructured content from the same domain. Research published in "REALM: RAG-Driven Enhancement of Multimodal Language Models" and related LLM retrieval studies consistently show that documents with clear factual structure, named entities, and explicit sourcing are retrieved and surfaced preferentially in RAG-based systems.

People's restructuring made its content more machine-readable. That's not a coincidence.

Why it worked: three structural reasons

Owned audience removes the intermediary. Every direct subscriber is a reader the AI search layer cannot intercept. Email and app-based distribution creates a parallel channel that compounds over time, independent of algorithm changes at Google, OpenAI, or Perplexity.

Structured content wins AI citation races. AI engines are retrieval systems. They surface content that is unambiguous, well-attributed, and factually discrete. When People tags an article with explicit author credentials, schema markup, and clear factual claims, it becomes a better candidate for AI citation than a competitor's wall of unstructured prose.

Brand trust transfers to new revenue models. The People brand existed before Google. It will exist after AI search. Licensing, events, and commerce deals are viable precisely because the brand carries meaning outside of search traffic. Most pure-digital publishers lack this asset. People has it. You can see a similar dynamic in how HubSpot has shifted from SEO-dependent blog traffic toward community, certification, and tool-based brand equity as AI erodes traditional inbound traffic.

What you can steal from this

The People case is translatable. The specific tactics differ by industry, but the structural logic holds for B2B SaaS, e-commerce, and professional services brands equally.

If you are watching your organic traffic erode and wondering whether zero-click search is temporary or structural, the Bloomberg framing is the honest answer: it is structural. The brands winning right now are not fighting it. They are routing around it.

HubSpot is the B2B parallel to People. It built an audience of millions on SEO-optimized blog content. As AI Overviews began answering "what is inbound marketing" and "how to write a cold email" without sending clicks, HubSpot's blog traffic declined materially. Their response, like People's, involved doubling down on owned tools (the free CRM, the website grader) and structured content that functions as AI-citable reference material rather than keyword-stuffed articles.

According to Semrush's State of Search 2024 report, branded search volume remains resilient even as generic informational queries shift to AI engines. Brands with strong direct recall, People, HubSpot, Spotify, Nike, are insulated from AI-intermediation in ways that commodity content sites are not.

The death spiral Bloomberg describes is real. But it applies most lethally to brands that built their entire identity on search-intermediated traffic with no direct audience, no brand licensing value, and no structured content worth citing. That is the trap to avoid.

Tracking your AI citation rate with a tool like winek.ai before and after implementing these changes gives you the baseline to know whether your structured content is actually being picked up by AI engines or just sitting there looking organized.

Your action plan

1. Audit your traffic source dependency , If more than 60% of your sessions come from organic search, you are overexposed to AI-intermediation risk. Estimated effort: 30 minutes in Google Analytics.

2. Restructure your top 20 articles with explicit schema markup , FAQ schema, HowTo schema, and Article schema with author credentials are the highest-leverage structural signals for AI citation. Estimated effort: 4 hours with a developer.

3. Build one owned audience channel this quarter , Email newsletter, push notification, or app-based delivery. Direct relationships compound; search referrals don't. Estimated effort: 2 weeks to launch a basic newsletter.

4. Add E-E-A-T signals to every author profile , Named authors with external publication credits, LinkedIn profiles, and domain-relevant credentials get cited by AI engines at higher rates than anonymous or thin profiles. Estimated effort: 1 hour per author.

5. Measure your AI citation baseline with winek.ai , You cannot optimize what you cannot measure. Run your top 10 brand queries across ChatGPT, Perplexity, Gemini, and Claude to see where you are being cited and where you are invisible. Estimated effort: 45 minutes.

6. Identify one brand asset that exists outside search , Licensing potential, certification programs, community, tools. If your brand has no equity outside of traffic, the AI death spiral hits you hardest. Estimated effort: Half-day strategy session.

7. Publish one definitive, structured reference piece per month , Not a blog post, a reference document. Named entities, explicit claims, sourced statistics, clear structure. The kind of content that actually drives AI recommendations is specific, not general. Estimated effort: 6 hours per piece.

Frequently asked questions

Q: What is the AI death spiral and why does it threaten publishers?

A: The AI death spiral describes the feedback loop where AI search engines answer queries using publisher content without sending traffic back to the source. Publishers lose advertising revenue, reduce content investment, and produce less high-quality material, which degrades the training data available to AI models over time. Researchers like Rand Fishkin and Rutgers professor Caitlin Petre have both flagged this as a structural economic threat to journalism and content creation.

Q: How are large publishers like People adapting to AI-driven traffic loss?

A: People Inc. CEO Neil Vogel has described a strategy that combines direct audience development (email, app, push), brand licensing deals, and content restructuring to make articles more AI-citable. The shift moves revenue dependence away from search-intermediated advertising toward owned channels and brand equity that persist regardless of algorithm changes.

Q: Does structured content actually get cited more often by AI engines?

A: Yes. Research on retrieval-augmented generation systems consistently shows that documents with explicit factual claims, named entities, clear attribution, and schema markup are retrieved and surfaced preferentially over unstructured prose. This is why FAQ schema, author credentials, and structured data are high-leverage investments for brands pursuing AI citation.

Q: Is the zero-click trend reversible, or is it structural?

A: Current evidence points strongly to structural. Google's own product roadmap, Perplexity's core business model, and ChatGPT's search features are all built on the premise of answering queries without requiring a click. Brands that treat this as a temporary fluctuation and wait for search referrals to recover are likely to be disappointed. The correct response is to adapt distribution strategy, not to wait.

Q: What is the B2B equivalent of People's adaptation strategy?

A: HubSpot is the clearest parallel. As AI eroded traffic to generic marketing how-to content, HubSpot invested in owned tools (free CRM, website grader), certification programs, and community that generate direct brand relationships independent of search. B2B SaaS brands can apply the same logic: build assets with direct user relationships rather than relying entirely on SEO-driven content traffic.

Q: How do I measure whether my content is being cited by AI engines?

A: Run your core brand and category queries across ChatGPT, Perplexity, Gemini, Claude, Grok, and DeepSeek and record whether your brand appears in the responses. Tools like winek.ai automate this measurement across engines and track citation rate changes over time, giving you a quantitative baseline before and after content restructuring efforts.