Boris Dzhingarov, founder of ESBO Ltd, leads a global Digital PR & SEO agency dedicated to enhancing brand awareness and visibility.
getty
Part of my job is deciding where coverage is worth earning, which means I look at more websites in a week than most people read in a year. A pattern shows up constantly. The site looks like a magazine: clean design, a familiar-sounding name, an archive going back a decade. Then you read three articles and check the numbers. The posts appear every 20 minutes, no author shows up twice, yesterday’s topics include pet insurance and industrial valves and the search traffic fell off a cliff a year ago. It is a publication-shaped object. Nobody is inside.
This matters to anyone who approves a PR or marketing budget, because the target lists those budgets pay for are full of these sites. The economics explain why. Producing articles has never been cheaper: Roughly half of the new articles published on the web are now primarily AI-generated, according to Graphite’s analysis of Common Crawl data. Building a readership, meanwhile, has never been harder. So the web now carries a thick layer of sites that publish endlessly and get read by almost no one. NewsGuard’s running count of AI content farms, news sites that operate with little to no human oversight, has passed 3,700 across 16 languages. Coverage on sites like these is a press release into a void, and in the campaigns I have tracked, it produces nothing measurable. The pages never even make it into a search index, let alone into an answer.
Five minutes of checking beats a month of pitching.
The good news is that these sites are easy to unmask, and none of the checks require special tools. Start by reading five recent articles. Real publications have authors who exist: a byline that appears regularly and belongs to a person you can find elsewhere, and writing that holds something a machine could not know, a quote the writer collected or a room they were in. Then look at what the site covers. A genuine trade publication stays in its lane because it serves a defined reader. A site that covers everything serves no one, and its breadth is usually the clearest confession.
Then check whether anyone, anywhere, treats the publication as real. Do people search for its name, and does any other outlet ever reference it? Social accounts settle it quickly, since real ones show conversation and hollow ones just broadcast. Traffic estimates from third-party tools can help, too, but read the direction before the number. A site whose visits collapsed after a search engine update has already received a verdict. And in 2026 there is one more check worth adding: Ask an AI assistant a few questions about the topics the site claims to cover, and see whether it ever gets cited.
I wrote here in April about how AI interfaces are reshaping the way buyers discover brands; one downstream effect is that assistants have become a rough census of which sources the wider web takes seriously. The census is lopsided. Graphite found that 82% of the articles cited by ChatGPT and Perplexity were written by humans. Finally, look for the masthead. A real publication tells you who edits it and how to reach them, and those editors have histories you can verify on LinkedIn in about two minutes. Hollow sites hide their people for the simple reason that there are none to show.
Use the arithmetic nobody likes.
Apply these filters honestly and a 500-site outreach list shrinks to 30. The 30 are slower to earn, and that is precisely the point. One article in a publication with living readers and working editors will outperform 20 that nobody, human or machine, will ever encounter. When a proposal promises coverage in impressive volume at friendly speed, volume and speed are the tell. Real editorial attention does not scale that way.
The filters do need judgment. A small trade journal with 5,000 readers can matter more than a large general site if those 5,000 are your buyers, so size alone proves little. New publications deserve a fair look, since every real outlet was new once. And detection cuts both ways: Plenty of legitimate newsrooms now use AI somewhere in their process, so treat a single suspicious article as a reason to read more closely rather than a reason to walk away. No single signal separates the real from the hollow. The question is whether evidence of an audience shows up anywhere at all.
The publication-shaped sites usually pass the metric checks. Authority scores can be manufactured. Readerships cannot. So before your team pitches anywhere, or before you approve an outreach list someone else built, spend five minutes inside each site’s archive and ask whether anyone is on the other side. The answer is usually obvious once you look.
Forbes Agency Council is an invitation-only community for executives in successful public relations, media strategy, creative and advertising agencies. Do I qualify?
