LinkedIn Post Formats: The 10 Types and How to Choose

The 10 LinkedIn post formats at a glance

A post format is the kind of post you pick in the composer, the box above the feed on LinkedIn's home page where you write. The feed is the stream of posts people scroll through. The format decides two things: what a reader can do with your post there, and what LinkedIn does with it after you publish. Members and company Pages (the accounts businesses post from) get the same formats, with a few thresholds noted below. The limits are LinkedIn's own, from its help pages.

Three other things look like formats without being one. A link turns into a preview card built from the page you link to. A repost shares someone else's post, with or without your comment. And the job and "find an expert" buttons belong to hiring and services, not to content. Image, document and video sizes have their own page: LinkedIn post size for every format.

Which post format reaches the most people on LinkedIn

Reach is the number of people who see a post; LinkedIn counts it as impressions, one per time a post appears on someone's screen. Engagement is what they do with it: reactions, comments and shares. Five companies that sell LinkedIn or social media software published format studies in 2026, each on the posts it tracks. Some report a median, the middle value: half the posts did better, half worse. None of the studies is a controlled test: the author, the topic and the effort change with the format, so these are tendencies, not laws.

The percentages can't be compared from one study to the next. Socialinsider counts reactions, comments and shares against impressions, while Buffer doesn't say what it counts on LinkedIn; a median of 21.77 % for documents suggests clicks are included, which LinkedIn's own Page analytics show next to reactions, comments and reposts. What each study shows is the order of formats inside its own sample.

The median matters because averages flatter formats with a few huge hits. In MyFeedIn's data a typical post gets 505 impressions, and video is the clearest case: its average is ten times its median, so most videos travel modestly and a few go very far.

The studies agree that documents sit near the top, for reach and even more for saves and comments: in MyFeedIn's data a document gets a median of 5 comments against 2 for a typical post, and in AuthoredUp's documents make up 12.92 % of all saved posts, about 2.6 times their share of what gets published. They also agree that video is the format whose typical result fell most in 2025 and 2026, and that reach is falling for every format (MyFeedIn's medians dropped for all of them between the two halves of its year). LinkedIn's engineers point to competition, writing in November 2025 that the number of posts created each day "has grown rapidly over the past year". They disagree on polls and on audience size, which has its own section below.

Documents' lead is often credited to dwell time, the seconds a reader stays on a post, as if the feed rewarded whatever takes longest to read. LinkedIn's feed engineers say they judge a long stay against posts of the same type, because "a dwell time of 30 seconds could be considered 'long' for an image but not for a video" (Fengyu Zhang, staff software engineer, September 2024), so no format is favoured for taking longer to read. A document gains when readers stay on it longer than on other documents, and when they react, comment or share.

LinkedIn itself refuses to rank. Dan Roth, its editor-in-chief and vice president of content, called video LinkedIn's "fastest growing content type" in an interview at Cannes published in July 2026, and said in the same answer that LinkedIn doesn't tell creators to post only video: the right format is the one that carries the message. He defined quality as a post that starts the right conversation, a meme included.

So use the data as a starting point, then pick by the action you want, and judge the post on whether that action happened: a document that gets saved, a poll that fills, comments that answer the question you asked.

If you have fewer than 5,000 followers

The format order changes for small accounts. AuthoredUp split a second, later dataset by audience size (personal profiles, October 2025 to March 2026), and its figures by band are these:

Polls' engagement stays between 0.16 and 0.27 times the usual post in every band. These band figures sit below the 1.78 times polls get in AuthoredUp's main table because they come from a different dataset: the main table covers all personal profiles from March 2025 to February 2026, the bands a later six months split four ways, and AuthoredUp gives no poll figure for the under-5,000 and 20,000-to-50,000 bands and doesn't reconcile the two sets. Read the main table as the overall order and the bands as the order by size.

Small accounts also earn more per reader: in AuthoredUp's data the band from 1,001 to 5,000 followers reached the top median engagement rate of all, 2.68 %, and it drops sharply above 50,000. Metricool found personal profiles 63 % more engaging than company Pages on average, and only 7 % of Pages grew enough in a year to move up one of its follower-size bands. For a company Page under 50,000 followers, Socialinsider's data favours posts with several photos for impressions.

So a freelancer with 800 followers should start with an image that shows real work and a caption that says what it proves, and keep documents for the ideas that need steps. Richard van der Blom, a consultant who has published research on LinkedIn's algorithm since 2020, advises small accounts not to compare their numbers with large creators': their audience and offer are different, and what counts is whether the right people start following or writing to you (May 2026).

Text and image posts

Text and images are the two formats most people already use: in MyFeedIn's sample, posts with one image make up 48 % of everything published and text alone 32 %. A text post holds up to 3,000 characters, a few hundred words; for anything longer, LinkedIn's help points to an article. The feed shows the first lines and hides the rest behind "…see more", and LinkedIn publishes no character count for that cut, so treat the first line as the label of the whole post.

Text posts reward length. In AuthoredUp's data (March 2025 to February 2026), text posts of 1,000 characters or more reached 1.18 times the format's average, and those under 300 characters 0.88 times. Plain text also does better for strong writers than the medians suggest: Jasmin Alić, a LinkedIn creator and writing coach, said on The Futur in December 2024 that his text-only posts still reached the most people, and van der Blom said in July 2026 that text-only posts were doing "much better than last year", with the warning that the writing then carries everything.

An image post carries one photo, or up to 20 in a multi-image post, and a post can hold a photo or a link, not both at once. Text with an image is what most people see: van der Blom estimates that 60 % of the posts shown on LinkedIn are text with a picture. For the caption, AuthoredUp found 800 to 900 characters the best length for image posts, with nothing gained beyond 1,000, and portrait photos earned more engagement than square or landscape ones in its sample of 368,489 single-image posts (August 2025 to July 2026). The image works when it carries proof (a result, a screenshot, a photo of the work); Dan Roth noted that people now show their work through posts because recruiters search for skills and want evidence. The first of several photos takes the most room and sets the layout. Slack's senior social media manager, speaking in Socialinsider's April 2026 webinar, described the team's use: several photos for event recaps, documents for anything that reads like a report. One detail matters if your readers find you through Google: LinkedIn says people who aren't logged in "might not be able to see certain types of posts, like multi-photo posts". The exact sizes, shapes and crops are on LinkedIn post size for every format.

Document posts, the LinkedIn "carousel"

What people call a LinkedIn carousel is a document post: you add a PDF, PowerPoint or Word file, and readers swipe through it page by page inside the feed. LinkedIn allows one document per post, up to 100 MB and 300 pages, asks for a title, and lets anyone who sees the post download the file as a PDF. Every page must have the same size, and LinkedIn advises converting to PDF before uploading.

It works for anything with steps: a framework, a checklist, a short report, a before-and-after. Each swipe keeps the reader on the post, and documents lead on saves and comments in the studies above. Short ones get saved: van der Blom observed in May 2026 that people save infographics and short documents, and that "people aren't going to save a carousel with 100 slides". AuthoredUp also found that short captions suit them: with a caption under 100 characters, documents reached 1.28 times the format's average. Once posted, the file can't be replaced; only the title and the description can be edited, so check every page on a phone first.

The name causes confusion because LinkedIn did have a real carousel. In July 2022 it launched a post of swipeable photos and videos, stopped letting people create them on 26 June 2023, and removed every one on 14 December 2023, according to its help page of the time. Today the only carousels on LinkedIn are document posts and the paid Carousel Ad.

Video posts

A video post holds one video, from 3 seconds (2 from the phone app) to 15 minutes, and up to 5 GB. A tap on a phone opens a full-screen vertical feed of videos, and in February 2025 LinkedIn brought a vertical video display to its website too. LinkedIn said then, as TechCrunch reported on 4 February 2025, that video uploads had grown 36 % in a year and video created on LinkedIn itself by 100 %; in July 2025 it reported uploads up more than 20 % over the year.

That growth has a cost for each video. AuthoredUp found that median video reach on personal profiles, March 2025 to February 2026, was 36 % lower than a year earlier. Socialinsider, in its April 2026 benchmark of company Pages, reported video views down 36 % year on year. That the three figures match is a coincidence: they measure uploads, reach and views, in different samples. MyFeedIn, for its part, puts the median video level with an ordinary post. Van der Blom said in July 2026 that LinkedIn boosted video from 2024 to mid-2025 while it tested its video feed, that video was back to normal proportions when his yearly paid report on LinkedIn's algorithm came out in April 2026, and that he expects a new push as the video tab leaves its test phase.

The advice and the data disagree on length. Louise Brogan, who runs a LinkedIn video agency, advised on the Social Media Marketing Podcast (June 2025) to keep videos below two minutes, which she believes is what LinkedIn wants. AuthoredUp's analysis of 36,946 videos (October 2025 to March 2026) found the opposite: videos over three minutes got 21 % more reach than the average video, and clips under 30 seconds the least. Two more things to know before comparing your numbers. According to van der Blom, a video view only counts once the video has played for about three seconds, while a post scrolled past already counts as an impression, so video numbers look smaller than they are. And video builds trust when you are at ease on camera; van der Blom's warning is that a video made only because the algorithm seems to want one shows the speaker's discomfort, and his advice to people who don't enjoy the camera is a newsletter or a text-and-image post instead.

Polls

A poll is a question with two to four answers (five from the iPhone app), open for one day, three days, one week or two weeks. The question takes up to 140 characters and each answer up to 30. Each member votes once, and nobody can make a poll anonymous: LinkedIn's API, the interface other software uses to publish on LinkedIn, rejects the setting.

Three rules make polls unlike any other post. The author sees how each person voted (for a company Page, its admins do), while everyone else sees only percentages. Your connections are notified that you created a poll. And a poll can't be edited once it's live, or reopened once it closes; it can only be deleted. LinkedIn also bars polls that ask for sensitive data, such as health, religious or political beliefs.

Across all the personal profiles in AuthoredUp's data (March 2025 to February 2026), polls have the widest reach of any format, 1.78 times the author's usual post, and the weakest engagement, 0.37 times. In AuthoredUp's later figures by audience size (in the table for small accounts above), the edge is smaller, and under 5,000 followers single images beat polls. For company Pages, Metricool and Socialinsider both found polls bringing the most impressions per post, Socialinsider only above 50,000 followers. A vote spreads the post without starting a conversation. Dan Roth, asked to choose between a poll and a strong opinion (a "hot take"), picked the opinion: polls take forever, he said, and he wants to know what the author thinks, not what everyone thinks. Van der Blom uses polls as the first post of a series instead: in his observation, people who vote then see the author's next post, so a poll on a question can be followed by an expert post on the answer.

Events and LinkedIn Live

An event is a page for something that happens at a set time: in person, online through LinkedIn Live, or online elsewhere. Creating one automatically shares a one-time post to your feed, and LinkedIn says the event isn't created until you publish that post. You can then invite your first-degree connections, the people you are directly connected to, up to 1,000 a week across all your events. Attendees get reminders seven days and three days before, and 15 minutes before the start, plus a notice if the time, place or status changes. LinkedIn also tells "a subset of the network of the organizers and speakers", so inviting a speaker with an audience widens the reach. LinkedIn's post scheduler doesn't take events: the event's post goes out the moment you create it.

What an event brings is a list. No published study measures how many people LinkedIn events draw, but the mechanism is documented: every attendee has raised a hand for a topic, and a company Page can add LinkedIn's registration form, which collects registrants' details for the organizer (personal profiles can't). Shanee Moret, who coaches business owners on LinkedIn Live, argued in a March 2026 livestream that this is the point: many followers read for years without reacting, and the attendee list shows who is actually paying attention. She recommends going live at least once a month.

LinkedIn Live is a live video stream inside an event, and its rules changed in 2026. Since 22 June 2026, "the ability to go live spontaneously is no longer available": every Live must first be set up as an event with a start time, even one only minutes away. Access goes to members and Pages with more than 150 followers or connections, an account at least 30 days old and a clean record under LinkedIn's community policies; there is no application. You stream through outside software (StreamYard, Restream, Switcher Studio, SocialLive or a custom setup), since LinkedIn can't broadcast directly, and when you go live the push notification goes only to the event's attendees, not to all your followers. LinkedIn did publish one comparison for Live: in 2019, comparing regular and live videos from the same broadcasters, its principal product manager for video, Peter Roybal, told TheWrap that live got 7 times more reactions and 24 times more comments. That was during the invitation-only beta; LinkedIn hasn't updated it.

Some ranking guides still say Live and newsletters require creator mode. LinkedIn removed the creator mode switch in March 2024; its tools stayed, and access to Live and newsletters now follows the rules above.

Articles and newsletters

An article is a long page published on LinkedIn: up to 125,000 characters, with images, links and text formatting, its own web address and, since 2024, a title and description of your choice for search engines (SEO settings). You write it from a computer; LinkedIn says the feature isn't available in the mobile app. You can change an article at any time, and the change is silent: your connections aren't notified, nothing marks it as edited, and the old version disappears. A lone article does poorly in the feed, though: 0.69 times the author's usual reach in AuthoredUp's data.

A newsletter is a series of articles that people subscribe to, and that delivery is what makes the difference. LinkedIn invites your connections and followers to subscribe when you start one; after each edition, it sends push, in-app and email notifications to every subscriber. Every member has been able to create one since mid-2024: in May 2024 LinkedIn's help still required more than 150 followers or connections, and by July it said all members could. A company Page needs more than 150 followers and a record of original posts, and an author can run up to five newsletters at a time.

Van der Blom gave his own numbers in July 2026: about 30,000 impressions per newsletter edition on 120,000 subscribers, against about 15,000 per ordinary post on 260,000 followers. That is one author's account, not a study. Dan Roth's advice for newsletters, from another 2026 interview: write every edition for someone reading it for the first time, and split it into takeaways, because readers scan.

Both formats also travel outside LinkedIn. For any post or article that anyone is allowed to see, LinkedIn's help says search services "such as Google, Bing, or ChatGPT" may show it in results or in AI-generated summaries.

Celebrations, links and reposts

A celebration is a templated post for a milestone, opened with the "Celebrate an occasion" button: welcoming a new colleague, giving kudos, a project launch, a work anniversary, a new position, a degree or a certification. It shows on your feed and your profile, with a LinkedIn illustration or your own photo. No study measures celebrations apart from other image posts, so the case for them is what they say: the news is the content, and it is news mainly to the people who know you. They are worth it for a new role, a launch or a certification, the proof of work Dan Roth described recruiters looking for, and for welcoming or thanking a colleague by name. For an idea or an opinion, a normal post does more.

Software that posts for you can't create a celebration: LinkedIn's API, the interface outside apps use, only lets them read celebrations.

A link post shares a web address, and LinkedIn draws a preview card from the page itself: its image, title and site name. A wrong or stale card is fixed on the website, then refreshed with LinkedIn Post Inspector. Where to put the link is the most argued question on LinkedIn. The evidence points two ways:

• A link in the post costs reach. Sprout Social tested it on its own company Page and published the result in April 2023: over a month and 8 posts, those with the link in the opening comment averaged 8,136 impressions against 3,309 for those with the link in the post. The two groups also had different text, so the test can't separate the two effects. Christopher Penn, a co-founder of the analytics firm Trust Insights, found that his own posts with a link got 38 % fewer impressions over a year, while noting that LinkedIn's engineering papers say nothing either way.
• A link left in the comments costs clicks. Van der Blom said in July 2026 that LinkedIn now hides comments containing links behind its "most relevant" sorting, so readers often can't find them.

LinkedIn's help documents no penalty for links. It does document one trap: a link added by editing a post later gets no preview card.

The choice follows the post's job. When the job is a click (a sign-up, a purchase, a download), put the link in the post and accept fewer impressions; that is van der Blom's rule. When the job is reach or a discussion, leave the link out of the post and offer it in a comment or on request, knowing fewer people will find it. Then check the clicks on your own site, as Penn suggests, rather than impressions alone: a post that reaches fewer people but sends more of them to you did its job.

A repost puts someone else's post in front of your network, with your comment ("Repost with your thoughts") or without. It is the weakest choice for your own reach, 0.29 times the author's usual reach in AuthoredUp's data, which advises writing your own post and tagging the author instead. In 2026 LinkedIn also began testing collaborative posts, with up to five collaborators and always public, open only to members of an early access program.

What you can change after posting

Most formats freeze when you publish, which is worth knowing before you click.

Scheduling works for most posts, from 10 minutes to 3 months ahead, but not for events; Scheduled post on LinkedIn covers it. Outside publishing software, such as ReadyToPost, goes through LinkedIn's API, which can publish text, images, video, documents, links and polls, but not celebrations.

If you meant formatting the text

"Format" also means the layout of the words inside a post, and Google mixes the two meanings for this search. LinkedIn's help describes no bold or italic for posts, and the text format of its posting API has no element for them: a post is plain text with line breaks, @mentions and hashtags. Articles and newsletters, written in a separate editor, do have bold, italics, lists and quotes.

The bold you see in some posts comes from websites that swap letters for lookalike Unicode characters. Unicode is the standard that numbers every character a computer can show, and its mathematics range holds symbols that only look like bold letters. Per Axbom, a designer who advises on accessibility, lists the cost: screen readers skip them or read each one out as a symbol name, search engines don't match them, and older devices may show question marks instead. Keep them, if at all, to a word or two, never your name or a full sentence.

What works without tricks: short paragraphs separated by an empty line, a first line that says what the post is about, numbered lines for steps, and a plain question at the end when you want answers.

Formats that no longer exist

Many guides disagree because they describe a LinkedIn that has changed. LinkedIn rarely announces removals, so the dates below come from its own help pages and posts, read in the Internet Archive's copies.

One idea, three formats

Here is a made-up example. A one-person bookkeeping service wants to remind clients of the three receipts they forget every quarter: parking, software subscriptions and the meal on a client visit. The idea stays the same; the format decides what the reader does.

Written as plain text, the first line is the label: "The three receipts my clients forget every quarter." The three items follow on three short lines, so they show on a phone before "…see more", and what counts as proof for each comes after "see more". The post uses a few hundred of its 3,000 characters. The sign it worked is a comment naming a fourth receipt.

Turned into a document, page one carries the same label in large type, pages two to four take one receipt each with the proof to keep, and page five says where to send them. Every page is exported at the same size, as LinkedIn requires, and viewers can download the file. The sign it worked is saves, and the check before posting is that page one reads as a cover on a phone.

As a poll, the question is "Which receipt do you forget most often?", 39 characters, the three receipts are the answers, each under 30 characters, and it stays open one week. The author sees who voted for what, which turns the result into a list of people to write to. Nothing can be fixed once it runs, so read the answers aloud before posting.

None of the three is the best format. They are three different requests to the reader: add a fourth receipt, keep the guide, or vote and become a name to write to. Decide which of those you want before you open the composer.