How to write LinkedIn posts in your own voice
By Jack Stovell X · Instagram · LinkedIn · published 2026-09-20 · updated 2026-09-30
LinkedIn has a voice of its own. One-line paragraphs. A hook that promises a lesson. "Here's what I learned." A list of three. A question at the end. It works, in the sense that the format is rewarded, and it has a cost: after a few months of it, your posts sound like everyone else's posts, and nothing like your emails, your newsletter or the way you talk in a meeting.
That gap is measurable, and measuring it is the first step to closing it.
See the gap. Paste something that is unmistakably you (a newsletter, a long email), then two or three LinkedIn posts. Each post is scored against the reference piece, with the habits that moved named and numbered.
What usually moves
Across the posts we have measured, the same features drift: sentence length collapses (the one-line paragraph), contractions drop (LinkedIn prose is oddly formal under the casual surface), questions multiply, and the argument structure flips to conclusion-first even for writers who build to a point everywhere else. None of that is wrong; it is the format. The problem is when the format replaces the voice rather than carrying it.
Keep the format, keep the voice
The two are separable. A LinkedIn post can be short and still have your rhythm, your contraction rate, your way of opening. The trick is to write from a target: the measured profile of your voice, with the post's structure as a constraint on top. That is exactly the brief ScriptGrain gives a model: 45 measured attributes plus the LinkedIn structure spec, then a score on the draft against the profile.
If you run LinkedIn for a team, the job looks different. We compare that setup with ScriptGrain in ScriptGrain vs Bloomberry. Bloomberry is built for LinkedIn distribution across a team: briefs in, approved posts out, under many names. ScriptGrain measures one person's voice and holds every draft to it, across sixteen content types, from blog to email to newsletter.
If you use an assistant for posts
Give it numbers, not "write a LinkedIn post in my voice". The prompt builder turns a sample of your writing into a prompt with your measurements in it; the connector hands ChatGPT or Claude the full profile and scores their drafts. Either way, check the post against the piece that sounds like you before it goes out. Under 0.55 and the format has won.
The measurements that matter most on LinkedIn
Not all 45 attributes pull equal weight in a post of 150 words. Some barely register at that length. Four do the heavy lifting.
Openers. The first line is the most exposed thing you write. LinkedIn rewards a hook, so most people reach for the same one: a bold claim, a confession, a number. If your emails open with context and your newsletter opens with a direct statement, a stacked hook is a voice change, full stop. Check your opening style and paragraph openers in the profile, then open the way you normally do.
Sentence-length variance. Average length matters less than the spread. You might average 17 words a sentence, but the real signature is the mix: a long sentence that builds, then a fragment. LinkedIn's one-line paragraphs flatten that to a steady drip of short lines. Variance is usually the first number to collapse, and the hardest to fake.
Contractions. AI posts are often oddly formal under a casual surface: "I do not think so" on a platform that's supposed to sound chatty. If you say "don't" everywhere else, say it here.
Line breaks. Fair enough, the format wants white space. You can have it. Break lines where your own paragraphs would break, not after every sentence. A paragraph of three sentences with one short one in it still reads fine on a phone.
A worked example
Here's an invented writer whose newsletter runs on contractions, mixed sentence lengths and a direct opener. Her first draft of a post looks like this:
I learned something important this year.
Consistency matters.
It is not glamorous. It is not fast.
But it works.
What has your experience been?
Every line is short. There are no contractions. The opener is a hook that promises a lesson, and the close is a question. It's the platform's voice, not hers.
Now the same idea, written from her profile with the format as a constraint:
Most of what I got right this year was boring. I turned up, I didn't skip the dull bits, and I kept a rhythm going even when nobody was clapping.
That's it. That's the lesson.
And to be fair, it took me far too long to stop looking for a cleverer one.
Same length, near enough. But there's a long sentence followed by a fragment, contractions throughout, a direct opening with no hook, and a dry sign-off in place of the engagement question. The format is still there: short, airy, easy to read on a phone. The voice is back in it.
Checking a post before you publish
Score the draft before it goes out. You get a voice-match figure from 0 to 1, and the useful part is the notes underneath: per-feature deltas, draft value against profile value.
Read the score first. On-voice drafts typically land between 0.85 and 0.95. A partial match sits somewhere from 0.5 to 0.8. Anything under 0.45 is a clearly different voice.
Then read the deltas, because they tell you what to fix. A quick pass:
- Opener: does the first line match how you normally start?
- Sentence-length variance: is the spread still there, or has it gone flat?
- Contraction rate: has it dropped against the profile?
- Paragraph shape: are the line breaks yours, or the platform's?
Polish can revise a draft toward a target score (the default is 0.9), which helps when one or two numbers have drifted. But treat it as a second opinion, not a substitute for reading the post aloud.
A post that scores 0.92 and a post that scores 0.6 can look identical at a glance. The difference only shows up in how it sounds, and the number gets you there faster than squinting at it. If the score comes back low, you already know which habit to go and find.
Questions
Should my LinkedIn posts score the same as my newsletter?
Not identically: a 150-word post and a 1,200-word newsletter differ in structure by design, and the score allows for that (short pieces cannot show sentence-length variance). A post that reads under 0.55 against your own writing has lost the voice, not just the length.
Can I measure a ghostwriter's posts against my own writing?
Yes, that is the founder-and-ghostwriter check: your own writing as the reference, their drafts as the comparisons. The features that moved are the notes to give them.
Does ScriptGrain write LinkedIn posts?
Yes: LinkedIn post is one of the sixteen content types, with its own structure spec, and every draft comes back with a voice-match score against your profile.