Creating content ·
LinkedIn hooks in 2026: how to open a post without sounding like an AI template
On LinkedIn, a post's first line decides whether somebody taps "see more" or keeps scrolling. It is the only fragment the feed always shows, and since 2024 it is also the part that fastest gives away a post written with ChatGPT: the same openings, the same turns of phrase and the same neutral cadence your own brain has already learned to dodge.
In 2026 the bar has risen. The hooks that worked two years ago ("Did you know that...?", "Here are 5 lessons I learned...", "Thread incoming") now cost you reach instead of earning it. This guide gives you the anatomy of a good hook, eight patterns that do pay off today, the burned out templates worth retiring and a three pass method for writing the first line after you have written the post.
By Sheena de PunkVoice · Edited by Mario Pérez
What a hook exactly is on LinkedIn
A hook is the first visible line of a LinkedIn post in the feed. On mobile it cuts off around 140 characters, on desktop around 210. Below it comes the "see more" and everything else lives behind that click. That fragment has a single job: buying from the reader the three seconds they need to decide whether the rest of the post deserves their time.
That work is not done by a headline, an image or a hashtag. It is done by a short sentence that has to do two things at once: promise something concrete the reader wants to read and demonstrate in its very form that there is a person behind it. If the promise sounds generic or the form smells of a template, the thumb keeps scrolling.
Confusing the hook with an article headline is the most common mistake for people coming from blogging or newsletters. On LinkedIn there is no headline. There is a person talking in a feed full of people talking, and their first sentence competes with twenty identical ones on the same screen.
Why the hooks of 2023 no longer work
Between 2020 and 2023 a dozen openings became standard and paid off handsomely: the rhetorical question, the surprising data point, the numbered promise, the shy confession, the "unpopular opinion". They worked because they were rare in a feed dominated by well written corporate posts. Once anybody with ChatGPT could generate them at scale, they became the feed's average register and lost the element of surprise holding them up.
On top of that comes generative pattern aversion: the perceptual rejection your brain triggers towards content with AI texture. In 2026, many readers identify a cloned hook in under a second and stop reading, even when the post behind it was written by a person. The first line contaminates the rest.
LinkedIn itself is adjusting its feed to reduce the visibility of generic AI-created content, and the most burned out patterns are the most visible signature of that content. A hook that smells of a template does not just repel human readers, it also depresses your algorithmic reach.
Generative pattern aversion: why your brain rejects content that smells like AI
The anatomy of a good hook in 2026
A hook that works today meets four conditions at once. They are the ones that show up systematically in the posts sustaining organic reach this year, and the ones almost always missing in those that fade after fifteen minutes.
- It is specific. It names a concrete object: a person, a number, a place, a date, a client, a scene. Generality ("many companies", "today's professionals") is AI's house style in cautious mode.
- It has tension. It announces a contrast, a broken promise, a half told scene or an opinion that can be argued with. A first sentence you could equally read or skip functions as a notice, and notices do not hook.
- It sounds like a person. It uses a spoken cadence, with inflections that only make sense in a specific person's voice. Active verbs, some personal lexical choice, some rhythmic asymmetry.
- It fits inside the feed's cut. It finishes the idea (or cuts the tension) before the "see more". A hook that needs to expand to make sense already lost the reader who did not tap.
Eight hook patterns that do work today
These are the eight patterns paying off in 2026, with concrete examples you can adapt. None is a rigid formula: they are ways of opening that still have enough life in them because they need material of your own or they fall apart.
1. The concrete scene
You land the reader in a specific moment, with an object and a verb. The camera switches on and something is already happening. Example: "Yesterday at quarter past eight, a client sent me a screenshot of one of our posts with a comment of his own: 'you could have written this with ChatGPT'".
It works because sustaining it requires a real memory. AI can invent the scene, but rarely with the minor details that make it sound lived.
2. The figure that does not add up
A concrete number that breaks expectations, low or high. Example: "This month we turned down 14 leads. All of them could pay".
It works because the figure opens a compulsory question: why? The reader cannot leave without the answer.
3. The opinion that could cost you clients
A position a cautious colleague would not publish. Example: "Hiring someone for their LinkedIn personal brand is, most of the time, a mistake".
Not to be confused with gratuitous provocation. The opinion has to be something you would defend face to face with whoever disputes it. If you would not defend it, do not publish it.
4. The operational confession
You tell a failure, a failed attempt or something you do not fully control. Example: "I have spent six weeks trying to close a client who says yes in every meeting and then does not sign. Here is what I am trying".
It works because in a feed full of successes, doubt is a rarity. And because an operational confession promises learning alongside you in real time.
5. The fragment of conversation
You open with a sentence somebody said to you or you said. Example: "A CFO told me this week: 'we are not putting AI in the team until we know what it is doing with the data'. And he is right".
Dialogue forces you to know where it came from and where it is going. It buys the three seconds almost on its own.
6. The asymmetrical comparison
Two things of the same kind performing very differently. Example: "Two posts published on the same day. One has 47 comments, the other 3. The only difference is the first two lines".
It works because it promises an explanation, and because the asymmetry rules out coincidence as the answer.
7. The closed operational question
No "Did you know that...?". A question with a concrete answer the reader can formulate in their head in three seconds. Example: "How long has it been since you updated your profile's About section? If it is more than six months, you are paying for it".
The implicit answer ("six months or more, definitely") makes the reader the protagonist and gives them a reason to continue.
8. The cut at the end of the paragraph
You cut the first line just before the key information. Example: "Two years ago I left a director role to set up a small consultancy. With what I learned this year, I would do it again, but I would change one thing".
It works because the "see more" promises something specific. Careful: only if you deliver behind it. If the post does not deliver that specific "thing", the reader feels cheated and does not come back.
Burned out templates: the seven openings that cost you today
These seven openings smell of AI from a metre away and trigger the reader's pattern aversion in the first second. They appear both in posts generated with ChatGPT and in posts written by people who internalised the feed's average register. In either case, today they penalise you.
- "Did you know that...?" followed by a generic data point. The feed's most exhausted rhetorical question.
- "In an increasingly..." A filler formula announcing that the post will say nothing concrete.
- "Here are X lessons I learned in Y years". The numbered promise burned out by ten thousand identical posts.
- "After working with over X clients / interviewing over Y people / analysing over Z cases...". The numerical appeal to authority nobody believes any more and AI repeats on autopilot.
- "Thread incoming" / "Save this post". Meta-instructions to the reader that assume their interest before earning it.
- "Unpopular opinion:". A label announcing what is coming and defusing the surprise. Besides, in 2026 almost no opinion arriving flagged that way is genuinely unpopular.
- "Everyone talks about X, but nobody talks about Y". A prefabricated antithetical formula AI generates as standard and that now works as a label of origin.
Write the hook after the post: the three pass method
The most common process mistake is trying to write the hook first, when you do not yet know what post you are going to write. It comes out forced, and usually lands on a template because the brain reaches for what it memorised from the feed. The rule that works best in 2026 is the reverse: write the whole post first and work on the hook at the end, once you know exactly what you are promising.
This is the three pass method you can apply to any post, whether it takes one minute or twenty.
Pass 1: write the post without thinking about the hook
Draft the whole body. Do not stop at the first sentence. Leave whatever provisional opening comes out, even if it is "this post is about X". The goal is having the real material in front of you before you start seducing the reader.
Pass 2: underline the liveliest scene, figure or sentence in the post
Reread what you have and mark the most specific, most loaded fragment: a concrete moment, a surprising figure, something somebody said, an asymmetrical comparison, a confession. That is the material your hook will be born from. Never pull it out of thin air. Always from the body of the post.
Pass 3: write three different hooks and pick the one that fits a template worst
Draft three alternative first lines using different patterns (a scene, a figure and an opinion, for example). Read them out loud. Discard the two that would sound the same written by somebody else and keep the one only you could sign. If all three sound alike, keep rewriting: it is a sign you are still inside the feed's average register.
A quick checklist before publishing
Before you hit publish, run your first line through these five questions. If you answer no to any of the five, it is not ready yet.
- Does my hook name something concrete (a person, a figure, a scene, an object, a date) that would not appear in a generic post on the same topic?
- Does the sentence open a tension, a promise or a question that forces the reader to continue to resolve it?
- Would it sound odd in another person's voice in my field, or could anybody publish it by copying and pasting?
- Does it fit entirely before the mobile "see more" (around 140 characters) without leaving the idea hanging?
- If I delete the hook and leave only the rest of the post, does the post lose something real, or was the hook decorative?
The first line is the smallest and most decisive editorial gesture
Hooks work as the smallest and at the same time most decisive editorial gesture of your personal brand on LinkedIn: the only sentence everyone reads before deciding whether to read you. Working on them well is the opposite of learning new templates. It consists of publishing sentences only specific people writing from a specific place could have written.
In a feed where the average register is mass produced by AI, any first line that sounds specific, carries tension and has a voice of its own becomes a useful anomaly. That is the whole secret and there is no shortcut. The advantage of having understood it in 2026 is that the bar is extremely low: most of the feed still opens exactly as it did in 2023.