Skip to main content
All posts
X Hooks

How to Write Hooks That Stop the Scroll on Twitter

A strong X hook names the reader, shows a tension or number, and makes the response obvious. Patterns from 9,556 posts and 10 posts with 2M impressions.

DS

Daniel Smidstrup

Founder of ClimbX

Follow on X

I grew my X account from 0 to 10K followers in 140 days.

5 min read
How to Write Hooks That Stop the Scroll on Twitter

A hook that stops the scroll on X does three things in its first line: it tells the right reader the post is for them, it shows a tension, gap or number they did not expect, and it makes the next step obvious, whether that is reading on or replying. Clever wording matters less than being specific.

What is a hook on X?

The hook is the part of a post a reader sees before deciding to stop. For a short post that is often the whole post.

For a long post or a thread it is the opening line or two, above the point where X cuts the text off. Everything after the hook only gets read if the hook worked.

A hook is not a trick. X's ranking model predicts negative actions as well as positive ones, including "not interested," mutes, blocks and reports, per its open-source code. A hook that promises something the post does not deliver earns exactly those reactions.

What does the data say about openings?

ClimbX's analysis of 9,556 posts from 185 creators scored every post against its own author's history. A post counted as a hit when it landed in that author's top 25%. Several findings apply directly to hooks:

  • Naming the audience works. Posts that opened by addressing a specific group, such as "Founders, here is...", hit 34.9% of the time.
  • Concrete stakes work best. Goal shares with a real commitment, deadline or number hit 50.0%, the strongest format measured.
  • Numbers help. A specific number lifted hit rate 1.21x (29.2% vs 24.2%).
  • Plain one-liners did worst. They hit 18.2% of the time and flopped 34.0% of the time.

Six hook patterns worth reusing

The examples below are real posts from Daniel Smidstrup's account, taken from his 10 biggest X posts, which together drew 2,067,969 impressions and 11,052 replies.

1. Address the reader directly

"Founders, which is the best place to buy a domain?" followed by five named options. The first word filters the audience, and the named choices make replying take one word.

2. Show a familiar pattern with one gap

The largest post listed the AI models the USA and China have, then asked "Europe has?" Two complete lists set the pattern and the gap asks the reader to fill it. That post alone reached 1,358,843 impressions.

3. Put the reader at the center

"Tell me one thing you can do that CLAUDE cannot do yet" uses a famous product as the hook, but the subject is the reader's own value.

4. Name a contradiction people already feel

"AI was supposed to make us work less so why does it feel like everyone is working 10x more now?" The tension is recognizable before the question mark arrives.

5. Add a tight constraint

"I am a founder scare me with 1 word" and "sell me your startup in two words max" turn a reply prompt into a quick game. Constraints lower the cost of answering.

6. Lead with a real number

"First X payout: $828.77. Took me 81 days to get here from never using X before." Amount, time and starting point make the milestone concrete, and the screenshots proved it.

How to write a hook in four steps

  1. Write the point first. One sentence: what should the reader know, feel or do after this post? If you cannot write it, the post is not ready for a hook.
  2. Find the tension. What is surprising, contested or unresolved about the point? A number that contradicts an assumption, a tradeoff, a gap in a familiar list.
  3. Draft three openings. Try one that names the audience, one that leads with a number, and one that states the contradiction. Pick the one a stranger would understand fastest.
  4. Check the promise. Read the hook, then the body. If the body does not pay off exactly what the hook promised, rewrite the hook, not the body.

Illustrative before and after

These are sample rewrites to show the mechanics, not measured results.

Weak openingStronger openingWhat changed
"Some thoughts on pricing.""Founders, we raised our price from $19 to $29. Here is what happened to signups."Named reader, real numbers, open question
"You won't believe this tool.""The analytics tool I use has one setting I wish I had found a year earlier."Specific object instead of a vague tease
"Consistency matters.""I posted every day for 50 days. The days that grew the account were not the busy ones."Stakes and a contradiction

How long should a hook be?

As short as the point allows. In the 9,556-post study, breakout posts that did at least 1.46x their author's normal engagement skewed to 13 to 25 words in total.

For everyday posts, 26 to 100 words had the best hit rate, so a longer post can work if its first line earns the read. How long should a tweet be covers length in detail.

Hook mistakes that cost reach

  • Burying the point. Context and backstory before the claim lose the reader before the claim arrives.
  • Vague teases. "Here is a secret" says nothing a reader can judge.
  • Bait that the body cannot pay. It invites the negative feedback the ranking model predicts.
  • Leading with a link. Posts with a URL hit 24.6% vs 27.0% without in the same dataset.

ClimbX shows what is working for accounts slightly ahead of you in your niche and helps you draft in your own voice with an assistant called Cliff. You review and edit every draft before it is published or scheduled.

Sources

Frequently asked questions

What makes a good hook on X?

It names who the post is for, shows a specific tension or number, and makes the next step obvious. It must also be paid off by the body.

Should every post have a hook?

Every post that competes for strangers' attention needs an opening that earns the next second. A casual reply to a friend does not.

Do questions make good hooks?

Specific ones do. In the 9,556-post study, genuine questions were the only breakout format with more replies than likes. Vague "thoughts?" questions rarely work.

How do I test hooks?

Keep the topic and format similar across posts, change only the opening style, and compare each post to your own recent median over a few weeks.

Try the loop on your own cohort.

Pick three accounts you would like to be at in 12 months. ClimbX pulls their recent outliers, tags them, and drafts in your voice off what is currently working. Edit, ship, watch the loop tighten.

Read next