We ghostwrite for ~40 ecommerce founders. Across the last six months we have watched LinkedIn dwell time eat every other ranking signal. Likes, comments and reshares still matter — but they now look more like outputs of dwell than inputs to reach.
If your post does not hold the eye for 1.7 seconds in the preview pane and 14+ seconds on the expanded view, the algorithm shuts the door before the comments game begins.
This is the playbook we use to engineer dwell on every post we ship.
What Dwell Time Actually Is on LinkedIn in 2026
Dwell is two distinct numbers LinkedIn measures separately:
- Preview dwell — how long the post sits in the user's feed without scrolling past it. Counted in milliseconds. The threshold to qualify for distribution rounds 2 and 3 sits around 1.6-1.9 seconds in our sample.
- Expanded dwell — how long the user stays on the post after clicking "see more," opening a carousel, or starting a video. The threshold for first-degree network re-injection sits around 14 seconds, and the threshold for second-degree network expansion sits around 38 seconds.
LinkedIn does not publish these numbers. We back into them by running matched-pair tests on our own client roster — same author, same week, same audience, different post structure.
Why Likes and Comments Lost Their Crown
In 2023-2024 a post could ride to 30K impressions on a flurry of comments in the first 90 minutes. That stopped working around Q3 2025. Comment pods, follow-trains and engagement chains stopped clearing the threshold.
The reason is mechanical. Comments are easy to gameify. Dwell is not. A user either reads or they don't. LinkedIn now treats dwell as the primary "did this post deserve attention" signal, and treats engagement as a corroborating signal.
In our client sample, posts with the highest dwell rates outperform posts with the highest comment rates by 2.4-3.1x in impressions when matched for follower count and author authority. We have stopped optimizing comment count entirely. We optimize the first three lines and the structure of the body.
The 1.7-Second Preview Dwell Test
A LinkedIn post in feed shows 210 characters before the "...see more" cut. That is your dwell qualifier. If those 210 characters do not earn the click, you do not enter the next distribution round.
Three structures we use to win that 1.7 seconds:
Specific number + uncomfortable claim. "After 50,000 listings, the badge most agencies stack hurts CVR by 14%." The number forces the reader to slow down. The claim creates an open loop.
Receipts opener. "Last month a $6M brand fired their agency mid-call. The audit doc is in the comments." The reader does not know the punchline. They click.
Reversal opener. "Everyone says X. The 320-SKU dataset says the opposite." The reader has heard X. They want to know if their belief is wrong.
What does NOT work for preview dwell in 2026: storytelling openers ("It was 11pm on a Tuesday..."), question openers ("Have you ever wondered..."), or list openers ("Here are 7 things I wish I knew..."). All three score dwell rates 30-50% below average in our sample.
The 14-Second Expanded Dwell Threshold
Once the user clicks "see more," you have roughly four reading-speed seconds for every hundred words. So a 350-word post earns about 14 seconds of dwell from a reader who completes it. Most users do not complete. They scan.
Structure for scan-readers:
- Bold the first noun in each paragraph so the eye can ladder down the post
- Use 3-7 word paragraphs in the middle third of the post — these act as visual rests and keep the reader scrolling instead of bailing
- Put the most concrete number at the 40-60% mark of the post, not the end. That is where dwell drops off
- Avoid stacked bullet lists longer than 5 items. Lists past item 5 score sub-9-second dwell in our tests
Document Posts and Video Posts Run on Different Dwell Math
Document/carousel posts score the highest expanded dwell of any format — median 42 seconds in our client sample, versus 17 seconds for text-only. The reason is mechanical: each slide swipe is a separate engagement event. We push founders toward 1 carousel per week, minimum 7 slides.
Native video scores high preview dwell (autoplay) but volatile expanded dwell. A 90-second video that loses the viewer at 8 seconds scores worse than a 12-second video that holds them. We coach clients to ship vertical, captioned, and front-load the hook in the first 3 seconds.
Text posts are the most efficient on dwell-per-minute-of-creation-time. They are also the most fragile to opener weakness. A bad first line on a text post is a dead post.
How We Measure Dwell Without LinkedIn's API
LinkedIn does not expose dwell time to creators. We back into it three ways:
- Impressions-to-engagement ratio. A post with strong dwell shows impressions climbing 6-18 hours after publish without proportional new engagement. That gap is dwell working.
- Follow gain per impression. A post that earns 1 follow per 200 impressions has held attention. The benchmark in our cohort is 1 follow per 400-600 impressions.
- Profile views per impression. This is the cleanest dwell proxy. A post earning 1 profile view per 80 impressions has hooked readers hard. Below 1 per 250, the post failed dwell.
We track all three on a weekly dashboard for every client. The brands that watch dwell-proxies instead of likes are the ones who scaled from 2K to 12K followers in 90 days.
Five Things That Tank Dwell
In our last 240 posts shipped, the patterns that crashed dwell:
- Question openers more than 5 words long. The reader scrolls.
- Links above the fold. LinkedIn down-weights outbound links inside the dwell-eligibility window. Drop links in comments.
- Tagging more than 3 people. Tags add friction without adding dwell.
- Generic stock photos as the post image. Image dwell drops to near zero. Use a screenshot, a number on a colored background, or no image at all.
- Posting in the second 18 hours after a prior post. Cannibalizes your own dwell pool.
FAQ
Does LinkedIn count dwell on muted videos? Yes, with a haircut. Autoplay muted video earns preview dwell at roughly 0.6x the rate of an unmuted view, in our matched tests.
Should I write longer posts to earn more dwell? No. Dwell is about earning the reader's attention per second, not extending their commitment. 280-450 word text posts are our sweet spot. Past 600 words, expanded dwell drops 22% in our sample.
Do carousels still work if you do not have a designer? Yes — plain text-on-color carousels built in Canva score similar dwell to designed ones. The structure matters more than the polish. Slide 1 is the hook, slides 2-6 are the argument, slide 7 is the punchline.
If your LinkedIn is not earning the dwell it needs, the playbook is not "post more." It is "engineer the first 1.7 seconds and the first 14 seconds." That is the work.
We run that playbook for ~40 ecom founders. Reach out if you want us to run it for yours.