Most founders treat LinkedIn like a slot machine. They publish a post, pull the lever, and hope. When it stalls at 200 impressions, they blame the hour, the hashtags, or the platform. I did that for years, and I was wrong about all of it. None of that's the real cause.
I spent 3 years inside LinkedIn, much of it on Sales Navigator and the Social Selling Index. That work gave me a close view of the feed's ranking machinery. The models have one job, and it's to predict which posts a specific member will stop and read. Nothing in the design is built to hide your posts. Those models sharpen every week.
Since then I've run visibility programs with founders and executives, and the pattern is consistent. Comment-first engagement produced roughly $340K in attributed pipeline across those programs in 6 months. That's the pattern below. The LinkedIn algorithm rewards a small, stable set of behaviors, and once you know them you stop guessing.
TL;DR - The Five Things That Actually Decide Reach
1. The feed is a cascade of ranking models, not one algorithm. Every post survives or dies at each layer.
2. Dwell time is the quiet signal. Seconds spent reading move reach more than a like ever will.
3. Comments carry far more weight than likes, and comment threads carry more than isolated replies.
4. The first 60 to 90 minutes set the trajectory. That window is the one variable you fully control.
5. Hashtags, engagement bait, and copied posting times do little. Some of them quietly cost you reach.
How the LinkedIn Algorithm Works: A Cascade, Not One Ranking
Start by dropping the idea of one algorithm. LinkedIn's engineering work describes the feed as a stack of systems that run in sequence, where each layer narrows the field before the next one starts. Your post has to clear every gate, and most don't.
Think of it as a series of interviews. Each system asks a narrower question than the last, and a weak answer at any stage ends the process early. Here's the sequence, from the moment you hit publish.
Candidate Generation
When you hit publish, your post joins a pool that spans more than 1 billion members and millions of daily posts. The first system asks one question: is this relevant to anyone at all? It reads your network composition, your past engagement, and the baseline signals on the post. Weak posts stall here, and posting into an unfocused network is the fastest way to stall before you start.
First-Pass Ranking
Posts that clear the first gate enter a ranking layer trained on billions of member interactions. This layer scores your content against relevance models, and it reads who is engaging rather than only how many. A post with 100 likes from unrelated accounts scores below a post with a dozen comments from people in your buyer's industry. I've watched that exact swap decide reach inside an hour.
Deep Relevance Scoring
The next layer reads the content itself. Topic classification, sentiment, and conversation quality all enter the score. A thread where several people reply to each other gets pushed further than a post with dozens of likes and no replies. Dwell time lands here as a first-class signal, because the model can separate a post someone read from a post someone scrolled past.
Re-Ranking and Filtering
Even a strong post faces one more pass. Diversity rules stop your feed from filling with a single creator. Freshness rules keep the feed current. Quality guardrails catch bait, spam, and recycled content. A post can rank well and still get demoted because serving it would make the feed worse for the person viewing it. That final pass produces the feed each member sees.
Every layer rewards the same 3 things: attention held, relevance to a specific network, and conversation. No layer asks how clever the hook was.
The Ranking Signals That Decide Reach
Not every signal carries the same weight, and you don't need all of them. That's true across dozens of posts and inside the client programs I track. Signals tied to real human attention dominate, and signals that are cheap to fake barely register.
The pattern is easy to read. Dwell time, comment depth, and network relevance sit at the top, and each one is expensive to fake. Likes, posting frequency, and hashtags sit at the bottom. You can buy those cheaply, and that's why the model discounts them.
Dwell Time: The Signal You Can't Fake
Dwell time is the attention a post earns, measured in seconds, before a reader moves on or reacts. The feed treats it as proof that the content was worth the space. A post a member reads for 30 seconds and then likes beats a post they like without reading. The model can't be fooled.
That's why long-form insight keeps beating the quick-hit post over a month, even when the short post wins on day 1. I started writing 60-second reads on purpose. When I compared 30 days of my own posts, the ones with the highest dwell time also had the widest reach, and the effect compounded as the algorithm learned who to show them to.
The rule for founders is simple. Write the post you'd want to read slowly. If a reader can skim it in 2 seconds and get it, they'll skim it in 2 seconds and forget it.
Key Insight
Dwell time is the one metric you can't buy. It's also the one the feed weighs the heaviest, so substance beats tricks over any window longer than a week.
Comments Versus Likes: How Engagement Gets Weighted
A like is the cheapest action on LinkedIn, and the model treats it that way. A comment costs real thought and time, and it produces text the ranking systems can read. Depth matters as much as the fact, and a quick click doesn't count. A paragraph-length reply is worth roughly 3.5x a one-word response.
The strongest signal of all is a thread. When the author replies and draws another reply out of the commenter, you've built something the model can measure directly. I design posts to start conversations, not deliver monologues, and the reach gap doesn't stay small. The comment pipeline framework behind my prospecting runs on the same mechanic. The comment is the unit of relationship.
“A like tells the feed that someone walked past your post. A comment tells it that someone stopped and thought. The model rewards the second one because that is what your buyer does long before they buy.”
Koka Sexton
Content Formats: What the Feed Amplifies
Format sets how much attention a post can hold, and there's no faking attention. The native formats rank by what shows up in my own results and in client programs.
- Text with a single image. The workhorse. Strong for one clear point of view and one clear ask.
- Document posts and carousels. High dwell time, because the reader swipes through frames. Best when every frame carries one idea.
- Native video. Strong when the first 2 seconds earn the third. Weak when the post is a talking head with no promise.
- Polls. Cheap engagement, low depth. Useful for reading a room, rarely for building authority.
- Link posts. The reserved format, and worth its own section below.
Match the format to the job. A framework belongs in a carousel or a long text post. A take belongs in a short text post. A relationship belongs in the comments.
Outbound Links and the Reach Tax
LinkedIn prefers to keep members on the platform, so a link post carries a reach tax. There's a real reason, since a link sends the reader away and cuts the session short. I don't fight this. I plan around it.
The default is to put the full idea in the post and drop the link in the first comment. If a link has to sit in the body, accept a lower ceiling and post less that week. Never hide a link inside a post that pretends to be a story. Readers notice, and so does the model.
When a link does go out, point it at something that keeps the relationship moving, like a free personal brand audit that answers the question the post raised.
Hashtags in 2026: The Honest Answer
Hashtags once carried real discovery weight. That's mostly gone now. The feed leans on semantic understanding of the post itself, so it reads your words and your audience more than the tags you attach. Current guidance is blunt: use 3 to 5 relevant hashtags, and only to help topic classification.
I tested 30-tag posts against 3-tag posts across my own content. The 30-tag posts didn't reach further. Some reached less. A wall of tags is a tell that the post was written for the algorithm, and the algorithm is trained to spot it.
The First 60 to 90 Minutes
The window right after publishing decides most of a post's life. The system shows your post to a small slice of your network first, then reads how that slice reacts. It's a test you can't skip. Strong early engagement widens the audience. Weak engagement narrows it, and the post rarely recovers.
Timing matters, but not in the way the generic advice suggests. The right time is when your specific audience is awake and active, and your own analytics will tell you when that's true. The bigger lever is what you do inside that window. Reply to every comment, and leave 2 or 3 thoughtful comments on posts from people you want to reach. That work hands the model fresh reasons to keep the post moving.
I've watched a post with modest likes and a deep comment thread outrun a post with 10x the likes and an empty comment section. The thread is what buys the second wave.
Posting Cadence: How Often and When
Frequency matters less than consistency, and both matter far less than the shape of each post. I post 4 to 5 times a week, and I'd rather publish 3 strong pieces than 7 weak ones. The feed rewards the creator who shows up with something worth reading, not the one who shows up most.
- Posting 7 times a week just to stay visible
- Copying a best-time-to-post chart from a blog
- Tagging people with no reason to engage
- Chasing likes with broad, safe takes
- Publishing, then walking away for a day
- Posting 3 to 5 times a week with a point of view
- Posting when your own audience is active
- Replying to every comment inside the first hour
- Writing posts built to start a thread
- Feeding the post for its first 90 minutes
The cadence that works is the one you can hold for a quarter. You're after a rhythm, not a burst. I'd rather see a founder post 3 times a week for 12 weeks than 12 times a week for two.
Where Founders Quietly Lose Reach
After enough profile audits, the habits that bleed reach without the owner noticing start to repeat.
- They connect at random. A network with no shared interest sends weak relevance signals, and the model lowers the ceiling on every post.
- They post and leave. The first 90 minutes are the cheapest reach they will ever buy, and they spend them off the platform.
- They write for the algorithm instead of a person. Bait, fake questions, and recycled advice get caught by the quality filters.
- They treat comments as a chore. The reply is where the relationship and the reach both live. That's the engine behind the 90-Day Executive Visibility Flywheel at the center of our client work.
Your First 30 Days
The feed rewards habits you repeat, not one-time moves. Start like this.
- Write for dwell time. Before you publish, ask whether a busy buyer would read this slowly. If they'd skim it, rewrite it.
- Set your window and defend it. Post when your audience is active, then block the next 90 minutes to reply to every comment.
- Start conversations, not broadcasts. End most posts with a question that has more than one honest answer.
- Build a relevant network. Prune the connections that dilute your signal, and add people who share your buyer's world.
- Put the link in the comment. Protect the reach of the post and give the link a fair shot at the click.
- Track dwell and comment depth, not likes. Keep a simple log for 30 days and let the data set your cadence.
For the scoring side, the compound signal scoring method is how I decide which posts and which people deserve my time on any given day.
How the LinkedIn Algorithm Works: FAQ
Does the algorithm penalize posts with links?
Yes, in a limited way. A link post tends to reach less than the same idea posted as text, because the link pulls the reader off the platform. Put the link in the first comment and keep the post itself complete.
Do comments really matter more than likes?
Yes. Comments take real effort and produce text the model can read, so they carry more weight than likes. A paragraph-length comment counts for far more than a one-word reply, and a back-and-forth thread counts for most of all.
How many hashtags should you use on LinkedIn?
3 to 5, chosen for topic fit. Hashtags no longer carry meaningful discovery weight, and a wall of tags reads as a post built for the algorithm rather than for a reader.
How long does a post take to take off?
Roughly 70% of a post's reach trajectory is set in the first 60 minutes, and the curve largely resolves inside 24 hours. The early window is where your effort pays the most.
How often should you post?
3 to 5 times a week, sustained for a quarter, beats a burst of daily posting. Consistency lets the model learn who to show your work to.
What is dwell time?
The seconds a member spends with your post before moving on. The feed treats longer dwell time as evidence that the content was worth the space, which is why posts people actually read reach further.
Does the algorithm show your posts beyond your network?
Yes, once early engagement from your own network proves the post is worth it. Strong signals inside your network are what unlock reach beyond it.
Does editing a post after publishing hurt reach?
Small fixes are fine. Large rewrites hours later can muddy the signals the post already earned, so proofread before you publish and leave it alone after.
The One Thing to Remember
The feed is a matching system. It's built to connect the right content with the right reader at the right time. When your work serves your audience, the system carries it further. When you try to game it, the quality filters catch up. Spend your effort on what your buyers actually read and reply to, because that's the only signal that has ever compounded.
Ready to make the algorithm work for you?
The 90-Day Executive Visibility Program turns your point of view into a consistent LinkedIn presence that generates inbound from buyers, partners, and press.
Book a Call →