LinkedIn's feed algorithm is not public documentation. What is known about it comes from LinkedIn's published blog posts, statements from its product and engineering teams, and observable behavior at scale. This piece draws on that public information and the patterns we observe in how founder posts perform differently from each other, which is not the same as having proprietary access to the algorithm's weights.
With that caveat stated clearly: the shift toward dwell time as a primary ranking signal is one of the more consequential changes in the last two years, and it has practical implications for how founders should structure their posts that are worth understanding.
What Dwell Time Measures and Why It Changed
Early versions of LinkedIn's feed ranking relied heavily on engagement rate, specifically the ratio of reactions and comments to impressions. This produced a well-documented incentive structure: posts optimized for getting reactions, which meant posts that asked binary questions, made provocative statements, or told emotionally resonant stories. The engagement rate metric does not care whether the reader spent 30 seconds reading the post or half a second clicking like. Both events count equally.
Dwell time measures something different. It measures how long a viewer's screen stays on a piece of content before scrolling past. A viewer who reads a 600-word post in full is generating substantially more dwell time than a viewer who scans the first line and clicks like. LinkedIn began weighting dwell time more heavily because it correlates more strongly with user satisfaction: readers who find something genuinely interesting spend time with it, and that time signal is harder to game than a reaction click.
For a founder, the practical implication is that a post that is genuinely interesting to read performs better algorithmically now than it would have two or three years ago. Posts designed to capture a quick reaction are relatively less advantaged. This is broadly good news for founders who have substantive things to say and worse news for founders who were relying on engagement bait.
The First Two Lines and the "See More" Threshold
LinkedIn truncates posts in the feed after two to three lines, displaying a "See more" link. Whether a viewer clicks through to read the full post is a meaningful behavioral signal: it indicates that the first two or three lines were compelling enough to warrant continuing. Posts with high "See more" click rates tend to perform better in distribution because that click generates both dwell time (the reader is now committing to the full post) and a behavioral engagement signal.
This means the first two lines of a founder post function differently from the rest of the post. They are the hook, but not in the manipulative sense. They need to create a specific, credible reason to keep reading. The most reliable way to do this is to open with a concrete observation that is recognizable and slightly counterintuitive, something that makes a reader think "that's not how I would have described it, but I want to know why you're describing it this way."
Compare these two openings for the same post about enterprise sales cycles:
"Enterprise sales are long and require patience." is self-evident and produces no reason to read further.
"We closed a deal in 11 days with a 2,000-person company. Same product, same pricing, different entry point. The entry point was the problem." is specific, raises a question (what entry point?), and creates a reason to read the explanation.
The second opening is not more sensational. It is more specific. That specificity is what generates dwell time, because readers who are interested in enterprise sales now have a reason to commit to the full explanation.
Comment Velocity in the First 90 Minutes
LinkedIn's algorithm evaluates posts in a layered way. An initial distribution goes to a small fraction of first-degree connections. The engagement rate and dwell time in that initial window determine whether distribution expands to a larger audience. Comment velocity, specifically the rate at which comments arrive in the first 60 to 90 minutes after posting, is one of the clearer signals the algorithm reads in that initial evaluation window.
Comments require more effort than reactions and are more valuable as signals because they are harder to generate passively. A post that receives ten substantive comments in the first hour is performing differently in the algorithm's evaluation than a post that receives fifty reactions in the same window, because the comments indicate a level of engagement that reactions alone do not.
For founders, this creates a tactical question: should you actively try to drive early comments by asking a specific question in the post? The answer is: yes, but only if the question is genuine and the post is substantive enough that the question follows naturally from it. A post that ends with "What do you think?" as a transparent comment-farming technique generates low-quality comments and does not provide the signal quality the algorithm is looking for. A post that ends with a specific, debatable claim and invites disagreement tends to generate substantive comment threads that do provide that signal quality.
Connection Strength and Relevance Weighting
The algorithm also weights by connection strength. Posts from first-degree connections who you interact with regularly appear with higher frequency and priority than posts from first-degree connections who you have never engaged with. This means your most engaged followers are disproportionately important for your initial distribution window.
Practically: the early comments that matter most for algorithm amplification tend to come from people who already follow you closely and engage regularly. These are the people who will see the post earliest and whose engagement generates the strongest initial signal. Knowing who those people are and engaging back with them regularly (commenting on their posts, responding to their comments on yours) is a compounding investment in your distribution capability.
This is not a cynical observation about gaming the algorithm. It is a description of what happens when you build genuine relationships with people in your field on a professional network. The algorithm is rewarding the behavior that makes the network genuinely useful, which is mutual engagement between people with shared professional interests.
What the Algorithm Does Not Reward
Certain patterns that used to generate strong algorithmic performance have become less effective. External links in the post body (as opposed to in the comments) are still deprioritized: LinkedIn does not want to distribute content that sends its users to other platforms. This has been consistently observed behavior for several years and has not changed.
Repetitive content formats also appear to perform with diminishing returns over time. The algorithm seems to evaluate whether a post is different in format and approach from recent posts by the same author. A founder who posts the same three-bullet-point structure every week will see declining performance on that format even if the content is different, while a founder who varies post structure, alternating between narrative posts, single-observation posts, and structured analysis posts, sees more consistent distribution.
We are not saying you should optimize every post primarily for algorithmic performance. A post that is genuinely interesting and useful to a specific audience will tend to generate the behavioral signals the algorithm is looking for as a byproduct of being genuinely interesting. The reverse is not true: posts optimized for algorithm signals rarely produce genuine reader value, and the algorithm is increasingly good at distinguishing between them. For more on the post structures that generate substantive engagement, see the piece on what Singapore founders are doing differently and the deeper discussion of how we approach content planning for different posting cadences.
A Practical Note on Interpretation
Algorithm behavior on LinkedIn changes. What is described here reflects the observable patterns from 2024 through early 2026, but LinkedIn continues to iterate on its feed ranking. The dwell time emphasis is well-documented and has been consistent for long enough that it is reasonable to build a posting strategy around it. The specific weights and thresholds can change without notice.
The more durable principle, one that predates dwell time weighting and will outlast whatever comes next, is that posts which are genuinely specific and substantive attract the kind of reader attention that algorithms eventually figure out how to recognize. Writing for a specific audience with real observations is a strategy that ages well. Writing to game current algorithm behavior is a strategy that requires constant revision. The founders building durable LinkedIn audiences are mostly doing the former, even when they do not articulate it that way. You can read more about the production mechanics that make consistent substantive posting feasible in the piece on breaking the burst-and-vanish pattern.