How to write LinkedIn hooks, according to 12,988 posts

A LinkedIn hook is the first 200-odd characters of your post, the fragment visible before the “see more” cut. It is the only part most people will ever read, so it decides everything downstream. To work out how to write LinkedIn hooks that actually earn attention, we ranked 12,988 real posts by engagement rate and compared the top 10% against the bottom half. The answer is not length and it is not statistics. It is person and tension: winning hooks put a specific human in a specific moment and leave the loop open.

What counts as a LinkedIn hook?

The hook is not a genre of sentence. It is a physical constraint. LinkedIn renders a preview of your post in the feed and hides the rest behind a “see more” control. On desktop web that preview runs to roughly 210 characters. On the mobile app it is shorter, because the column is narrower and every hard line break you type consumes part of the budget before the character count does.

That last point gets ignored by almost every hooks guide, which discusses length in the abstract as though there were one universal cut. There is not. The same 205-character opener can show three full lines on a laptop and stop mid-sentence on a phone. Since most feed reading happens on mobile, the practical rule is stricter than the desktop number implies.

Our dataset captures each post's text up to the fold, which is a limitation if you want to study whole post bodies and a gift if you want to study hooks. Everything below describes the exact fragment a scroller judges. The rest of the full engagement study covers the format and metadata findings that sit outside the text.

What do 12,988 posts say about LinkedIn hooks?

We filtered a public dataset of 34,012 influencer posts down to 12,988 English posts from 65 creators, scored each one by (reactions + 4 x comments) divided by follower count, then split the ranking into a top decile of 1,298 posts and a bottom half of 6,494. Here is what the two groups do differently inside the visible hook.

Two of those rows are the interesting ones because they show nothing. Numbers in the hook split 32.1% to 32.8%, which is noise. Opening with a question word splits 12.4% to 10.6%, which is close to noise. Both are standard advice. Neither survives contact with the data.

The engagement gap those hooks are competing over is wider than most people assume. The median post in the whole cohort drew 122 reactions and 12 comments. The top-decile median was 235 reactions and 72 comments. The bottom-half median was 80 reactions and 5 comments. Reactions roughly triple between the bottom and the top. Comments go up more than fourteenfold.

Does hook length matter on LinkedIn?

No, and this is the finding that saves people the most wasted effort. The median visible hook measures 206 characters in the top decile and 205 in the bottom half. In words, 36 against 34. Whatever separates a hook that works from one that does not, it is not the number of characters you spent.

The advice you have probably read (“keep it under two lines”, “one sentence, maximum”) is not wrong so much as beside the point. A one-line hook and a three-line hook both work if the reader ends the fragment wanting the rest. Short openers get credit for a property they do not own: they tend to be punchy because writing something short forces you to cut the wind-up, not because shortness is doing the work. Cut the wind-up directly and the length takes care of itself.

There is one length rule worth keeping, and it is the mobile one above. Get the tension in early enough that a tighter crop cannot swallow it. If your best line is at character 190, half your audience never reaches it.

Which LinkedIn hook patterns actually win?

Six shapes show up repeatedly in top-decile posts. They are not templates so much as starting positions, and most strong hooks combine two.

1. The first-person moment

A specific thing that happened, told plainly, with the outcome held back. This is the pattern behind the biggest split in our data: 19.6% of top posts open with I, my, or we, against 10.1% of weak ones.

Yours does not need to be devastating. “Great first day as CEO” posts and “I got this completely wrong” posts sit in the same top decile. The requirement is a real moment, not a big one.

2. The direct address

Name the reader, or a group the reader belongs to, in the first line. Nearly half of top-decile hooks put the reader in the post somewhere.

The subversion matters as much as the address. “Dear recruiters, I know you have a difficult job” sets up a complaint. “I mostly don't mind” refuses to deliver it. That gap is the reason to keep reading.

3. The declaration

A flat, confident claim the reader either strongly agrees with or wants to argue about. Declarations are the most common top-decile style in our exemplar set, and the argument-starting kind earns comments, which our scoring weights four times over reactions.

The failure mode of a declaration is being unarguable. “Culture matters” is a declaration nobody will contest, which means nobody will comment. Push until the sentence has an edge somebody could push back on.

4. The real question

Questions over-index in top posts and over-index hardest among the comment magnets. The condition: it has to be a question you want answered, aimed at the reader's experience, not a rhetorical ramp into your own answer.

5. The pattern break

Set an expectation in the first clause and violate it in the second. This is the mechanism under a lot of hooks that get filed as “contrarian”, but it works without any contrarian posturing.

6. The repeated line

Anaphora, the same opening words repeated down a short list, turns a plain observation into rhythm. It also survives the fold well, because each line is complete and the reader can feel the pattern continuing below the cut.

More structural patterns, including how the body should carry a hook once you have one, are in the content strategy guide.

Why do first-person hooks outperform?

We can measure the correlation. The mechanism is inference, but there is a documented piece of the machinery worth knowing about. LinkedIn's engineering team has published on using dwell time as a feed ranking signal, including a model that predicts the probability a member skips an update. Dwell time is measured from the moment at least half of a post is visible. In other words, the platform is watching whether the fragment holds you, before you tap anything at all.

That reframes the hook. It is not only a click-through device for the “see more” link. It is the thing being measured while a reader hesitates. A topic sentence (“Delegation is a critical leadership skill”) is instantly legible, which means it is instantly skippable. A first-person moment is not resolvable at a glance, so the eye stays a beat longer. The pattern that reads as “more human” is also the pattern that is slower to process.

The honest caveat: our data cannot separate the opener from the author. Some accounts write first-person hooks and also have unusually engaged audiences. The finding survives when we hold the author constant, which is reassuring but not conclusive.

Should a LinkedIn hook open with a statistic?

Almost every guide says yes. Our data says it makes no difference. A digit appears in 32.1% of top-decile hooks and 32.8% of bottom-half hooks. If leading with a number were doing work, that gap would not be pointing the wrong way.

The plausible explanation: a statistic in line one is a claim about the world, and claims about the world are easy to agree with and move on from. Numbers earn attention when they are weird enough to demand an explanation.

Compare that with a number used as narrative compression, which is a different move entirely.

Do emojis belong in a LinkedIn hook?

Emoji in the visible hook show the widest ratio in the entire study: 24.3% of top-decile posts against 2.7% of bottom-half posts. Nine to one. It is also the finding to hold most loosely.

Emoji travel with a persona. The creator who drops a rocket into line two is usually also writing casually, in first person, about something that happened to them. So a chunk of this effect is the first-person finding wearing a costume. The defensible version: an emoji will not hurt you, and the register that refuses them entirely is the same register that produces bottom-half posts. If emoji feel wrong in your voice, do not force them. Fix the person and the tension instead.

Do these hook patterns hold across topics?

Partly, and the split is instructive. We divided the cohort into 20 content topics and compared each topic's own top decile against all of its posts. The reader-addressing effect travels almost everywhere. The first-person effect is choppier.

Look at career growth. First-person openers actually tick down in its top decile, while second-person ones climb nineteen points. Career advice is a genre where the reader is the subject, so the winning move is to point the sentence at them rather than at yourself. Startups runs the opposite way: first person more than doubles and second person barely moves, because a founder story is only interesting if it is yours.

The practical version: pick the pronoun your topic rewards, and check where your category already sits before assuming you need to change anything. Per-topic breakdowns with real hooks are on the post examples hub, and the role-level cuts sit on the founders and CEOs page.

Should the hook change if the post is a video or an image?

Almost nobody segments hook style by media format, which is odd, because the person writing the caption for a video is solving a different problem from the person writing a text post. Here is how the visible hook actually differs by format across the whole cohort.

Video captions are written in a different register entirely. They ask questions at nearly 1.6 times the rate of text posts, use emoji three times as often, and address the reader more. That makes sense once you picture the surface: a video in the feed autoplays silently, so the caption has to do the work of a trailer rather than a story. It has to say who this is for and why the next thirty seconds are worth unmuting.

Shared-article posts sit at the bottom of every column, which is the same shape as the format finding in the pillar study. The caption on a link post is usually a headline restated, so it addresses nobody and asks nothing. That is a writing problem, not a format problem, and it is fixable: see do external links hurt LinkedIn reach for what a link post that does work looks like.

Document posts, the format behind carousels, show the most extreme hook profile in the data, with 58.3% using “you”, but there are only 60 of them across 13 authors. That is far too small to build a rule on. We mention it because it is suggestive and because most pages would quietly round it into a headline.

How do you rewrite a weak hook?

Most bad hooks are not badly written. They are correctly written summaries of a post, placed where a hook should be. The fix is usually subtraction. Three worked examples follow. These are invented for illustration, not drawn from the dataset.

Rewrite 1: remove the wind-up

Before:“Over the past few years I have thought a lot about hiring, and I wanted to share some reflections on what I have learned about interviewing engineers at a small company.”

After:“We hired the candidate who failed our technical screen. She is now our best engineer.”

The first version is a table of contents. The second is the thing itself. Notice the second version does not resolve: the reader still does not know why she failed, or what changed.

Rewrite 2: put a person where the topic is

Before:“Effective delegation is one of the most under-rated skills in modern management.”

After:“I stopped reviewing my team's work for a month. Two things broke. One of them was me.”

Same claim, same post underneath. The first has nobody in it, so there is nothing to be curious about. The second has a decision, a consequence, and an admission.

Rewrite 3: move the tension forward

Before:“Last quarter we ran an experiment with our pricing page, testing several variants over six weeks with a reasonable sample, and the results genuinely surprised the whole team.”

After:“Raising our price by 40% increased signups. I still cannot fully explain why.”

The first version buries its own payload at character 150, past the tightest mobile crop. The second lands the surprise in the first eight words and adds an admission of confusion, which is an open loop a reader can help close in the comments. More on turning that loop into an actual conversation is in how to end a LinkedIn post.

How do you pressure-test a hook before publishing?

1. Cut it at 140 characters, not 210. If the tension survives the tighter crop, it survives every surface. See how to start a LinkedIn post for the opener-by-opener version of this check.
2. Find the person.Is anyone in the frame, or is it a topic sentence? “Delegation is a critical leadership skill” has nobody in it. “I stopped reviewing my team's work for a month” does.
3. Find the open loop. If the hook resolves itself, the reader has no reason to tap. Hold the outcome, the number, or the lesson until after the fold.
4. Check who could have written it. If a competitor could publish your hook verbatim, it contains no information. Add the detail only you have.
5. Ask what a comment would say. If you cannot draft a plausible reply yourself, the post will get reactions at best. Comments were the widest gap in our data.
6. Read it out loud. Top hooks read like speech. If it sounds like a press release, rewrite it the way you would tell a friend over coffee.
7. Delete your first sentence. Then read it again. Most drafts start one sentence before the story does, and the deletion is usually free.

What ruins a LinkedIn hook?

• Throat-clearing. Any opener whose job is to describe the general state of modern business is a scroll-past instruction. Start at the moment instead.
• A link in the visible hook. Link posts under-perform, and in the pre-fold fragment they also spend your best real estate on something the reader cannot act on yet.
• A hashtag wall. Four or more tags is a bottom-half signal, appearing in 41.9% of weak posts against 25.9% of top ones. The hashtag data is blunt about it.
• A forced statistic. The data says it buys nothing on its own.
• Announcing the format.“A thread on...” or “Here are 7 lessons...” tells the reader what shape the post is before giving them a reason to care about its contents.
• Front-loaded credentials. Your title is on your profile. Spending the first line establishing standing costs you the only line that has to earn the tap.

How do you write good hooks consistently?

Individual good hooks are not hard. Producing one every working day is, and that is where most people quietly stop posting. The approach that survives contact with a real schedule is to separate collecting from writing.

1. Keep a running note of moments, not ideas. A sentence about something that happened, captured the day it happened. Ideas go stale. Moments do not.
2. Write five hooks for one post. Hooks are cheap and bodies are expensive. Draft five openers, pick one, and let the winner decide what the post is about.
3. Steal structure, never content. Note the shape of a hook that stopped you, then fill it with your own specifics.
4. Track comments, not reactions. Comments were the widest gap between the top decile and the bottom half in our data, and they are the metric a hook can actually move.
5. Publish on a schedule you can hold. Frequency compounds only if it is sustainable. See how often to post and when to post.

What this analysis cannot tell you

These are correlations, not causes. We can say top hooks open in first person about twice as often as weak ones. We cannot prove the opener produced the engagement. The cohort skews toward established creators, all with at least 1,000 followers across 65 accounts, so treat the figures as directional for a new profile. Engagement counts are lifetime totals at the time of collection and post ages vary, which adds noise to any single post's score. And every text finding describes the visible hook, not the full body, because that is what the source data captures. If you want the complete method and the myths it killed, it is all in the 34,000-post study.