The trend filling your LinkedIn feed this month, the algorithm rants, the Substack discovery complaints, the essays asserting the machines have lastly gained, is constructed on actual numbers pointed on the mistaken conclusion. That’s the argument Tony Uphoff made in his Substack submit. (Uphoff and I each labored at Ziff-Davis between 1988 and 1993. Even then, each platform shift in commerce publishing obtained handled by any individual within the business as the tip of the enterprise.)
In response to Uphoff’s Substack submit, natural clicks have fallen roughly 42% since Google’s AI Overviews scaled throughout search outcomes, a determine Outline Media Group reached by pulling Search Console information throughout 64 websites and one which matches an older Pew Analysis panel discovering that solely 8% of searchers click on by means of to an natural consequence when an AI Overview seems, towards 15% when it doesn’t.
LinkedIn natural attain is down 47% year-over-year for the reason that platform rebuilt its feed rating round 360Brew, a 150-billion-parameter mannequin that reads a submit’s experience and dwell time as a substitute of counting reactions. A Pangram Labs scan of greater than 1,000,000 posts discovered 41% of LinkedIn’s long-form content material is now absolutely AI-written, the best share of any platform measured, which is a part of why LinkedIn added a “looks as if AI slop” reporting button on July 30.
Then there’s the workforce quantity. Mercer’s 2026 World Expertise Developments survey put worker worry of shedding a job to AI at 40% worldwide, up from 28% two years in the past. Challenger, Grey & Christmas has tracked AI as the explanation cited in additional than 101,000 introduced job cuts by means of the primary half of 2026, roughly double the full-year 2025 tempo.
Line up the dates, and Uphoff has a degree that none of this appears to be like gradual. Substack overhauled Notes discovery late final yr. Google’s AI Overviews went from a small share of queries at launch to roughly half of them by this spring. LinkedIn shipped 360Brew that very same quarter. Three unrelated firms moved in the identical six-month window, which is precisely why it doesn’t really feel like coincidence to the individuals dwelling by means of it.
Structural Disruption Is Not The Similar Drawback As Decline
Uphoff’s precise argument isn’t about any of these platforms particularly. It’s a few distinction he thinks the business retains lacking. Decline means the factor you’re doing has no model price saving, but structural disruption means the perform nonetheless issues, and the shape it takes is altering beneath you.
He’s lived by means of that hole earlier than, at ThomasNet when print directories went digital and at UBM TechWeb when the 2008 monetary disaster hit the identical month because the print-to-digital collapse. In each circumstances, the loudest response within the constructing wasn’t technique; it was individuals indignant on the channel that was disappearing, not the guess the enterprise had made on staying tied to it. Uphoff’s closing line is the one price highlighting in daring: “The machines aren’t taking something from you that you just haven’t already agreed to lease out.”
Search Engine Methods conferences within the 2000s have been full of individuals satisfied Google’s subsequent algorithm replace would kill their enterprise mannequin. A few of these companies did die. Those that survived stopped arguing with the algorithm and began asking what perform they really carried out for a reader or a purchaser, then rebuilt round that perform as a substitute of the channel. I believe it’s the piece practitioners skip previous quickest as a result of arguing with the algorithm feels extra like doing one thing than rebuilding does.
The place I’d Push Again
Right here’s the place I would like extra precision than Uphoff’s column provides. Not each quantity in his personal “receipts” part survived contact with my scrutiny.
The 85% Enterprise Insider visitors collapse Uphoff cites got here from a Wall Road Journal chart that the Journal itself corrected on July 22, three weeks earlier than his piece ran. The corrected trailing 12-month figures put Enterprise Insider’s decline at 43%, not 85%, alongside a 28% drop for USA Immediately and a 44% drop for The Washington Publish. The unique chart had already circulated extensively throughout the website positioning business, together with in posts that drew tens of 1000’s of views, earlier than the correction caught as much as it.
That doesn’t undercut Uphoff’s argument. A 43% visitors loss nonetheless helps “that is structural, not decline.” But it surely’s a reminder that AI-era numbers transfer quick sufficient that the model you noticed shared final month could already be stale, and citing a corrected determine with out noting the correction is precisely the form of factor that erodes belief in each different quantity within the piece.
| Declare within the supply column | Authentic supply | Verification standing |
| Google natural clicks down 42% since AI Overviews scaled | Outline Media Group evaluation of Search Console information, 64 websites | Confirmed, present as of March 2026 |
| Natural CTR 8% with AI Overview current vs. 15% with out | Pew Analysis Middle panel research | Confirmed |
| Enterprise Insider Google referral visitors down greater than 85% | Wall Road Journal, authentic chart | Corrected by WSJ on July 22, 2026, to 43% |
| LinkedIn natural attain down roughly 50% yr over yr | A number of analyses of LinkedIn’s 360Brew rollout | Confirmed, constant throughout impartial trackers |
| 41% of LinkedIn long-form posts absolutely AI-written | Pangram Labs research, 1M+ posts, April to June 2026 | Confirmed; Pangram itself calls the determine a decrease sure |
| Worker worry of AI job loss at 40%, up from 28% | Mercer 2026 World Expertise Developments survey | Confirmed |
| 101,000+ AI-attributed job cuts by means of mid-2026 | Challenger, Grey & Christmas month-to-month reviews | Confirmed at 101,743 by means of June 2026 |
What To Really Do About It
Uphoff’s column ends with a intestine test as a substitute of a method, which is a good alternative for a management essay however leaves website positioning practitioners with out the following step. Right here’s my model constructed for this viewers.
First, separate your channel metrics out of your perform metrics. Observe natural clicks and LinkedIn attain if you need, however construct your actual dashboard round whether or not AI methods cite your content material, whether or not electronic mail subscribers convert, and whether or not direct visitors to owned pages is rising. These are the numbers that inform you if the perform you carry out is undamaged even whereas the previous channel numbers fall.
Second, audit what you really personal. E-mail lists, first-party information, and direct shopper relationships don’t get repriced by an algorithm replace or a report button. If most of your visibility nonetheless is determined by a single platform’s temper, that’s a enterprise danger to repair this quarter, not a speaking level to your subsequent submit.
Third, cease measuring content material by whether or not it ranks and begin measuring it by whether or not it earns a quotation, a save, or a reply. LinkedIn’s personal algorithm now rewards saves and sends over reactions. Google’s AI Overviews reward content material structured for extraction and named sourcing. Writing for the previous sign in a system constructed on the brand new one is the quickest method to mistake structural disruption for decline.
My view is easier than Uphoff’s essay makes it sound. Each platform talked about on this column modified what it rewards this yr. None of them stopped rewarding worth. The practitioners who spend the remainder of 2026 proving they nonetheless ship it, on no matter channel really carries it, are going to look so much smarter 12 months from now than those nonetheless arguing with a feed algorithm within the feedback.
Extra Assets:
Featured Picture: Roman Samborskyi/Shutterstock
