Between April and June 2026, 30% of public LinkedIn comments were written entirely by AI, according to detection startup Pangram. On September 18, LinkedIn answered: comments are no longer ranked by when they were posted, but by how relevant they are to the person reading. If your company's LinkedIn playbook ran on being first in the thread with a warm two-word reply, that playbook was retired last week without a memo.

Read it as a feed tweak and you miss the point. It is the platform formally admitting that the cheapest trust signal in B2B marketing has been inflated into worthlessness.

Why did LinkedIn change how comments are ranked?

LinkedIn switched to relevance ranking because chronological order rewarded the fastest, emptiest reply, and a growing share of those replies were machine-written. Pangram's analysis, reported by Gizmodo in July, flagged 41% of LinkedIn's long-form posts and 30% of short-form content as fully AI-generated, the highest of any platform it measured; Medium came in at 31%, Reddit at 13%, Substack at 10%. Originality.ai's separate scan of 5,000 long-form LinkedIn posts in July 2026 put the "likely AI" share at 81.2%. The methods differ, so the numbers don't stack. The direction is not in dispute.

The sequence tells you how the platform thinks. First came a "seems like AI slop" report button at the end of July, with chief product officer Hari Srinivasan telling Fortune the company had blocked billions of automation attempts in a couple of months. Then the ranking change. Blocking bots is expensive; making them invisible is cheap.

Why does a cheap signal stop meaning anything?

A signal carries information only while it costs something to send; once the cost falls to zero, everyone can send it, so it tells the receiver nothing. Michael Spence's 1973 job-market signalling model and Amotz Zahavi's handicap principle in biology land in the same place: the peacock's tail is honest precisely because it is a burden.

For a decade, a LinkedIn comment was that kind of tail. A senior person in your sector stopping to write two sentences under your post meant they had spent time on you, and Robert Cialdini's social proof did the rest: if people like that are paying attention, so should I. Generative AI removed the burden. And when nine of thirty comments are synthetic, the reader cannot tell which nine, so they discount all thirty. The real comments pay for the fake ones.

Effort was the currency of social proof. Automate the effort and you start printing your own money, and you are the first one to pay for the inflation.

What does this do to B2B buying decisions?

It hits them directly, because B2B buyers judge vendors on thought leadership more than on sales collateral, and that thought leadership is now being read with suspicion. In the 2025 Edelman and LinkedIn B2B Thought Leadership Impact Report, 64% of buyers said they trust a company's thought leadership more than its product sheets when assessing capability, and 71% of "hidden buyers", the stakeholders who never take a sales call, rated it more effective than conventional marketing.

For brands in Turkey and the wider region this is not a distant US story. DataReportal counts 21 million LinkedIn members in Turkey at the end of 2025, with the ad audience growing by a million between July and October alone. A bigger audience also means more of the same prompt-shaped posts landing in front of that hidden buyer as one grey pile.

Which budget line just got repriced?

Anything priced on volume just got repriced: engagement pods, bulk posting packages, services that comment on behalf of executives by the hour. Under chronological ranking the logic was simple; the first reply sat on top and the top reply earned visibility. Under relevance ranking, being first buys nothing.

Take a hypothetical company buying 25 support comments on each of 40 posts a month, 1,000 comments in total. Yesterday their visibility depended on timing. Tomorrow it depends on how well each commenter matches the reader's professional graph. Most of those 1,000 will never reach anyone's first screen. The same money could fund ten substantive replies from people who actually work in your category.

Effort, meanwhile, is regaining its premium. According to figures reported by The Sauce, time spent in LinkedIn comment sections rose 18% year on year. Attention is moving into the replies. It is just not moving toward empty ones.

What should you do in the next 30 days?

Shift your LinkedIn spend from volume toward effort that can be verified. Specifically:

• Reread any engagement or comment-service contract this month. Every clause that defines delivery as a count is now pricing a depreciating asset.
• Let AI draft executive posts, but require each one to carry something only that person could know: a number from the business, an observation from a client meeting, a mistake they made. Machines can't fake testimony.
• Replace likes as the success metric with a list of who commented, by title and company. A dozen of the right decision-makers is worth more pipeline than hundreds of anonymous reactions.
• Once a week, read the comments under your own posts and count the template phrases. If that share is rising, your audience quality is falling, and spotting it early costs nothing.
• Skip the "Great insight!" race entirely. Its prize was withdrawn on September 18.

The question is no longer how visible you are on LinkedIn. It is whether the reader believes a human's effort sits behind what they see.