Bloomberg Just Put a Face on the AI Search Collapse
By Matt Kott | Modern AISEO in AI search has collapsed 60-80% in the past two years.
On Bloomberg’s Wall Street Week, David Westin sat down with Neil Vogel, CEO of People Inc., the largest online and print publisher in America. 175 million readers a month. People, Better Homes and Gardens, Food and Wine, Entertainment Weekly. Vogel told Westin something that every brand team should stop and read twice.
“Our audience on the internet was 75% from Google Search, 2022 … That is now 25% of our audience.”
The shift happened over about four years. His business is profitable and growing. He saw it early, moved fast, and built new distribution before the floor gave way.
Watch the Bloomberg segment here.
Most brands are not People Inc. They do not have Vogel’s scale or his five-year head start. What they share with him is the same underlying problem he just described on national television: the gateway to your customer has moved, and most of the tools brands use to measure visibility are still watching the old door.
Why This Is Not a Publisher Story
This is not a media-industry story. Vogel’s traffic numbers are the most visible leading indicator of a structural change happening across every category.
Rand Fishkin, whose clickstream research tracks the full arc of this shift, gave Westin the longer view: “Whereas in 2011, Google was sending north of 70% of all searches to a third-party website… today that number is in the mid-40s and falling.”
Caitlin Petre, a Rutgers professor and author of “All the News That’s Fit to Click,” added the consequence: “News sites that were counting on referrals from Google for their traffic, for their ad revenue… are now seeing, in some cases, very dramatic decreases in the traffic that they’re getting from Google.”
The data behind that collapse, 863,000 keywords measured by Ahrefs and a second read from BrightEdge, is in our Q2 State of AI Search Report. The short version: in early 2024, roughly 76% of pages in Google’s top 10 were also cited by AI engines. By February 2026, that overlap had dropped to 17-38%. Two independent measurements, same direction.
The collapse is documented. What Bloomberg added is something different: a named executive at scale, on camera, describing the exact shift in plain language.
What Vogel’s Numbers Actually Tell You
People Inc. could see the shift because the traffic volume made it impossible to miss. A decline from 75% of 175 million readers shows up in revenue before it shows up in a dashboard.
Most brands do not have that signal. They have a Google Analytics account that looks stable. They have keyword rankings that have not moved. They have an organic traffic line that is holding. None of those instruments measure what Vogel is describing.
The gateway moved. The analytics stack did not.
Your customers are typing questions into ChatGPT, Claude, Gemini, and Perplexity. Those platforms generate answers. Recommendations. Brand comparisons. “What should I buy for X” queries that used to end with a Google results page now end with a synthesized paragraph from a language model. And the brand that appears in that paragraph is not determined by your Google rank.
The same shift is now happening inside Google itself. AI Mode and AI Overviews now surface synthesized brand answers directly inside Google Search, as a separate, parallel surface from traditional results. If you have searched Google lately, you have probably run into them. Most brand teams are not measuring them.
Vogel gave the strategic frame for why that matters: “AI needs three things. It needs power … It needs a model … And it needs inputs. We are the inputs. We are the training of the models.”
He also described what People Inc. did about it. Last summer, they partnered with Cloudflare to block every AI crawler that would not pay for access. They treated their content as a licensed input with a price, not a free resource for model training.
“We license content to LLMs and AI engines,” Vogel said. “It’s real money. It’s material for us.”
Most brands cannot license their content at that scale. But the strategic logic underneath it applies directly: if your brand is not a well-represented, authoritative input in AI training data, it is invisible in the AI answer. And if it is invisible in the answer, the traffic consequence is already in motion. You just cannot see it yet, because your dashboard is measuring clicks, not citations.
The prescription, documented in detail in our AI Visibility Intelligence framework, is measurement first. You cannot close a gap you cannot see.
Two Brands in Every Category Right Now
Twelve months from now, every category will have separated into two groups.
The first group is measuring AI visibility today. They know which query types their brand appears in, where competitors are landing instead, and what the gap looks like broken down by AI engine. They are closing it.
The second group is not measuring it. They are assuming their SEO position reflects their AI position. The assumption worked in 2021. It does not work now.
Vogel warned where this goes if publishers cannot sustain the content economics: “This death spiral… has been something that people have theorized over the last three or four years with no discernible solution.” His optimistic scenario for his own business is conditional on having brands strong enough to survive the traffic collapse and diversify. Most brands do not have that buffer.
The measurement gap is where the work starts. The full prescription is in our Measurement Gap post, including what the correct instrument looks like and what the data shows when you run it.
What You Can Do Right Now
Start with the public Try. Enter your brand once. We run four reads in parallel: how AI sees your brand in search, how authoritative sources represent you, how customers experience your brand, and how AI systems understand your business. No call required, no email required to see results.
The first read is the one this post is about: your Shortlist Share, the query categories where competitors are named instead of you, and the gap broken down by AI engine.
If the scores are what you expect and there are no surprises, you have confirmation and you are done. If they show gaps your current stack cannot explain, that is the conversation worth having.
When you book a call, we come already knowing your brand and your top competitors. If we are a fit, we set you up with a full diagnostic in the app, yours to keep, buy or not.
One question worth sitting with: if your three most important competitors ran AI visibility checks this week, what do you think they would find?
This post cites industry research from Ahrefs, BrightEdge, and Bloomberg segment appearances. Statistics are attributed to their named sources. Modern AI has not conducted independent replication of third-party studies. Individual brand visibility outcomes vary by category, query type, model, and measurement period.