LinkedIn is testing two tabs at the top of the feed: For You and Network.
For You is the default. It's the feed you have today, ranked by the algorithm, strangers included. Network only shows posts from people you've connected with or chosen to follow.
Hari Srinivasan, their Chief Product Officer, announced the test last Friday. His framing was that people want more real connection with real people, and this is a small part of rebuilding that.
My feed filled up with takes within a day.
Where I land, plus everything else LinkedIn shipped since I last wrote one of these.
1. This is the fourth time
2022: a Discover feed with topic tabs. Gone.
February 2025: All and My Network tabs, a five-week test. Gone.
June 2025: For You, Following and News pills on iOS. Pulled by August.
June 2026: topic-based suggested feeds. I flagged them in the July issue.
September 2026: For You and Network.
Facebook ran the same experiment in 2018 across six countries and killed it. Adam Mosseri's summary was that people told them they didn't want two feeds, were less satisfied with what they saw, and didn't connect more with friends because of it.
LinkedIn hasn't published ranking details for either tab, and none of the previous tests made it to a full rollout.
What I'd do: change nothing about your strategy this week. Treat the test as a signal about where LinkedIn wants to go, and keep posting the way that's been working.
2. Your post still gets ranked the same way
The Network tab is a filter on the feed you already have. Same posts, same ranking model, minus the strangers.
Your path into For You didn't change either. 360Brew still reads your first 50 words, matches the topic to people who care about it, and carries your five author data fields with the post. Everything in the April algorithm issue still applies.
And you've been able to see this split for months. The discovery section in post analytics breaks reach into in-network and out-of-network. I covered that metric in June. That number was already telling you how much of your audience is people who chose you versus people the algorithm sent.
What I'd do: pull your last 10 posts and note the in-network vs out-of-network percentage on each. If Network becomes a permanent tab, in-network percentages could increase. Out-of-network is the reach you earn on every post with these strategies.
3. If it sticks, followers get a more important job
I've been saying for a year that follower count stopped deciding reach. LinkedIn's own test data showed the rebuilt system lifted results for members with fewer connections, and Shield's (RIP) data on 50,000 posts showed a top post from a 5 to 10K account beating a median post from a 25 to 50K account. Both are in the April breakdown.
That's still true inside For You.
A Network tab adds one thing: a feed where the people who chose you see your posts without competing with strangers for the slot. Your connection list becomes a distribution list inside the platform.
Which doesn’t changes is who you should be connecting with. Every random accept is a person in that feed who will never buy.
What I'd do: send connection requests to the 20 to 30 people you'd most want reading every post. Clients, prospects, the creators you're already commenting on.
Keep this in mind for the next point too, because connections now matter in two places.
4. Comments are now ranked per viewer
This one landed in August and got buried under the slop news.
Comments under a post used to be one ordered list everybody saw. Now each viewer gets their own order, based on their professional interests, who they're connected to, and what they engage with. The comment at the top depends on who's looking.
same post, both of us are mostly just seeing our connections comments’
LinkedIn said time spent in comments is up 18% year over year, and a third-party analysis from Pangram found 30% of comments between April and June were fully AI-generated. So they had a comment section people were reading more and trusting less.
Two things follow from this.
Commenting first stopped buying anything. The old play was to get in early, collect likes, and sit at the top for everyone who arrived later. There is no “top” anymore.
And the comment your ideal client sees first is the one from someone they're connected to or someone in their lane. Which is the same argument as point 3: the connections you have with the right people now decide whether your comments get seen, as well as whether your posts do.
What I'd do: keep the engagement list at 20 to 30+ creators whose comment sections your buyers read. And connect with those people you want seeing you.
In July I told you LinkedIn was pulling the AI writer out of the composer and replacing it with something that only proofreads. It's in my composer now.
Post Proofreader is Premium only. It shows suggested edits with inline change tracking, like tracked changes in a doc, and you accept or reject each one. LinkedIn's wording is that it “improves clarity without changing your voice”.
Wow, surpriseeeed about the distinction they're drawing. Proofreading is assistance. Generating the post is slop. lol. Who would have thought.
The only AI tool I trust to not write AI slop is kleo.so or my own Claude setup.
But they've now built that line into the product.
What I'd do: you can use it for last minute check typos and nothing else. If it ever suggests rewriting a sentence into something you wouldn’t say out loud, reject it.
LinkedIn published numbers on September 15.
→ More than a million members used "Seems like AI slop" in the first two weeks
→ Views of content classified as slop are down 40% (idk Rick…)
→ The button now works on comments, not only posts
→ If your post gets enough flags, the tip shows up in your post analytics.
That was announced in July. It's live now.
They also repeated the automation numbers: hundreds of thousands of automated comment attempts caught daily, billions of automation attempts blocked in the last couple of months.
A million reports in two weeks tells you how much people wanted this. The 40% tells you the classifiers are working on more than the flagged posts.
What I'd do: nothing new if your posts already have a client situation, a number, or an opinion (that is actually YOURS) in them, even if you do use AI. If you've been pasting raw AI output, you now have a metric that will tell you.
7. LinkedIn wrote its own AI citation playbook
On August 11 LinkedIn published a guide for B2B marketers on getting cited by ChatGPT and the other chatbots. They're calling the framework the Credibility Stack.
Two things in it worth pulling out.
Long-form articles on LinkedIn attract more AI citations than feed posts.
And freshness counts, so a piece from 2024 loses to one from last month.
I wrote about this in the AI citations issue in June, and this is LinkedIn confirming the same thing from their side. The platform is the second most cited domain in AI search, and they know it.
What I'd do: if you've been treating LinkedIn articles as a dead format, reconsider them for the pieces you want cited. Reach is still better on posts. Citations are better on articles. Different jobs.
The pattern, again
Every one of these points the same direction as the summer batch.
A feed of people you chose.
Comments ranked by who you know.
Proofreading instead of generating.
A million people reporting slop and a 40% drop in views for it.
Credibility as the thing AI models look for.
LinkedIn is rebuilding the platform around the question of who is behind the words.
Which means the least scalable thing you own, your actual network of real buyers and the specifics only you have, is now the thing protecting your reach in the feed, in the comments, and in AI answers.
On the feed test specifically, Dani Markovits made the point that it comes down to whether people open either tab again tomorrow.
My bet is most won't.
Facebook tried the same split in 2018 and killed it because people didn't open the second feed.
We say we want to see the people we know, then the moment they give us the choice we drift back to the algorithmic feed because it's more fun and it's built for the part of our brain we don't control. People on here love complaining about the TikTok style For You feed, but that's what we pick when it's up to us, so my guess is most of us end up living in For You and forgetting the Network tab is even there...
Talk soon,
D.
PS: If you've got the two tabs, reply and tell me which one you opened this morning. I want to know before LinkedIn tells us.