Why a comment on LinkedIn now matters more than the post

On 9 August 2026 LinkedIn, the social network for work and professional connections, changed the order of comments under posts. Before, they were sorted by time: whoever wrote first sat higher. Now each reader sees their own thread, assembled around their professional interests, connections and activity. For anyone promoting themselves or a B2B company, meaning a business that sells to other businesses, this boils down to one simple thing. A smart comment under someone else's post now earns impressions on its own, separately from the post. Below I'll walk through what changed, why LinkedIn wants this, and where ChatGPT fits in.

How LinkedIn now sorts comments under a post

The first person to notice the change was Lindsey Gamble, a creator economy expert, who wrote about it on Threads. Then Social Media Today, an industry publication covering social platforms, gave it a detailed breakdown. On 10 September 2026 the agency FINN Partners recapped the news in its monthly roundup. According to LinkedIn's description, a post now shows the comments that suit a specific person. To pick them, the platform looks at that person's professional interests, their connections and what they usually engage with.

In practice, two people open the same post and see different top comments. A marketer sees a reply from a marketer in their network, a finance person sees a reply from a finance person. The rule "get your comment in within the first 10 minutes" no longer works. The one who rises to the top is the one whose reply is closest to the reader's interests. Speed decides nothing. What decides is how much substance the reply has.

Why LinkedIn did this: time spent in comments grew 18%

LinkedIn states the goal plainly: more timely discussions in the feed, so people talk more inside the app. The numbers give it a reason. According to LinkedIn's Q2 2026 report, time spent in comments grew 18% year over year. Overall content consumption over the same period grew 10%. So comments are growing 1.8 times faster than everything else. The platform is betting where its time is growing.

There's an awkward detail. The startup Pangram Labs analyzed 57,000 public LinkedIn posts. By its estimate, from April to June 2026, 30% of comments were entirely generated by AI. Pangram sells an AI detector, so the figure deserves a discount for the seller's interest. LinkedIn itself gave no exact statistics. Social Media Today suggests that a noticeable share of the comment growth comes from exactly these tools.

So whose growth is that 18%? Partly, it seems, it belongs to bots writing "thanks for sharing" to each other. Which brings us to the second half of the changes.

How LinkedIn is cutting AI slop in comments

AI slop is the name for text a neural network churns out in bulk, with no thought and no author behind it. LinkedIn has been fighting it for months, and in the comment section you can see it step by step.

1. August 2025reduced visibility for comments from automation tools
2. February 2026stepped up the fight against comment spam
3. May 2026LinkedIn learns to recognize signs of AI slop
4. 2 August 2026test of a "looks like AI slop" report button
5. 9 August 2026comments ranked by relevance

The report button lives in the three-dot menu on a post or comment. LinkedIn's director of product Hari Srinivasan explained it this way: AI slop is hard to define, the definition keeps shifting, and people's reports help tune the models. Authors were promised private flags in their analytics, showing which posts readers considered inauthentic or written with heavy AI help. Social Media Today also reports on measures against engagement pods, groups whose members agree to like and comment on each other's posts.

There's no ban on AI. Srinivasan said it directly: if AI helps you sharpen an idea, keep using it. Authenticity should be judged by real people, not by a detector that can get it wrong. Tellingly, LinkedIn scaled back its own generative AI. The "enhance your post" feature used to rewrite a draft. Now a proofreader has replaced it: it fixes mistakes and leaves the author's voice alone. Social Media Today reads this as a quiet admission that the platform's AI tools were hurting engagement.

Two more details. Since April 2026 a post's author can pin replies from verified accounts to the top. More than 100 million users have already verified their identity on LinkedIn. FINN titled its September roundup "Platforms Trade Reach for Trust." In the agency's view, LinkedIn treats genuine conversation as a ranking signal and protects it.

What kind of LinkedIn comment gets impressions now

LinkedIn has shown impressions for your comments since February 2025, and under any post, your own or someone else's. That means the comment has long been a separate unit you can measure. Now it has its own algorithm too. The three signals in LinkedIn's description, interests, connections and activity, hint at what will rise:

• A comment on your topic under a post from your industry. It will be shown to people with the same professional interests.
• A comment where your connections are reading. The connections signal works in favor of someone the reader knows.
• A reply with substance: a number, a case from your work, disagreement with an argument. "Great post" is exactly what that 30% of bots write, and it's the first thing to get hit by the report button.

An unpleasant truth for anyone who put commenting on autopilot through automation services. Since August 2025 LinkedIn can reduce the visibility of such comments. Since August 2026 anyone can report them with one click. 100 template replies a day no longer build reach; they damage the account's reputation.

Why LinkedIn matters for visibility in ChatGPT and Perplexity

ChatGPT Search, Perplexity and Google AI Mode answer a question with finished text and show which pages they relied on. In March 2026 HeyOrca, a social media scheduling service, summarized a study: LinkedIn ranks 2nd by citations across these three systems and appears in 11% of answers on average. That's roughly 1 in 9 answers. HeyOrca doesn't give the methodology or sample size, so I'd treat the figure as a guide, not a law.

LinkedIn itself is pushing in this direction. In February 2026 the platform published tips on optimizing posts for AI chatbots. In August 2026 it released a playbook for B2B marketers on visibility in AI search. For a company selling a complex service to other companies, the logic is simple. A buyer asks an AI who can solve their problem. The answer may be assembled partly from LinkedIn. Whether AI cites comments specifically, rather than posts and profiles, nobody has shown yet. But an expert comment brings people to your profile, and a profile with posts already shows up in answers.

What else changed on LinkedIn for creators and brands in 2026

• Collaborative posts, June 2026. A post can be published by several co-authors at once, and brand settings let you control which collaboration requests reach you at all.
• A personalized recommendations feed, in testing since June 2026. Before that, in March 2025, LinkedIn tested a My Network tab: only content from your contacts and people you follow.
• A new feed algorithm, 12 March 2026. What to show and in what order is now decided by more advanced AI systems.
• Live streams, April 2026. LinkedIn removed native live broadcasts.
• Frequency. LinkedIn recommends 2 to 5 posts a week, 1 or 2 of them video.

A separate note on BrandLink, because there's confusion around it. It's an advertising product for brands, not monetization a creator switches on themselves. Its predecessor, Wire, launched in June 2024: brands placed pre-roll ads, meaning a clip before videos from Bloomberg, Business Insider, Forbes and The Wall Street Journal. On 1 May 2025 Wire was renamed BrandLink and ads started running on videos from selected creators as well. Those creators get a share of ad revenue; how much, LinkedIn hasn't disclosed. As of September 2026 BrandLink hasn't expanded.

What to do

1. Open the analytics for your comments and see which ones earned impressions over the past month. That's your baseline before the new sorting.
2. Pick 10 authors in your industry that your clients read. Reply under their posts with substance, a number or a case from your work, 3 or 4 times a week.
3. Drop any services that comment on your behalf. After August 2026 they work against your account.
4. Keep 2 posts a week on your profile, so someone who arrives from a comment has something to read and AI search has something to cite.
5. If you have an industry partner, make one collaborative post together and see whose audience responds more strongly.