How to Get More Replies on X: Post Shapes That Start Conversations

This guide is about replies to your own posts. Replying to other people is a separate growth tactic with its own rules, covered in the reply guy strategy. Here the question is simpler: what makes someone stop scrolling and type an answer under your post?

Two limits before you copy this. Short posts win on my account, not on everyone’s. Across about 54,000 original posts from 669 ClimbX users (last 180 days), posts of 46 words or more reached 1.14x their author’s median and posts of 1 to 8 words 0.92x, so test length against your own median. And replies measure reach and conversation, not business. A post that draws 800 one-word answers rarely brings a qualified lead or a sale. For that you still need posts that show what you know: specific results, stories, how you work.

Why do replies matter more than likes on X?

Because X has historically ranked posts by the conversation they are expected to start. The heavy ranker X open-sourced, with weights dated April 5, 2023, weighted the predicted chance of a reply at 13.5 and of a like at 0.5, and the chance that the author engages with a reply at 75. Those weights apply to predicted probabilities, not to a points tally of real interactions, and the README says they can be adjusted at any time. Read them as history. The direction is the useful part: the model valued conversation far more than likes.

On my own account, across 1,230 original posts from the last 180 days, the median post drew 51 replies, or 17.7 per 1,000 impressions. The numbers in this guide come from four samples, each labelled where it is used: my own 1,230 original posts on @DanielSmidstrup, a label study of my top and bottom posts, a study of 500 outlier posts from 181 creators, and anonymized aggregates across ClimbX users. How each one was measured is at the end.

Do questions get more replies on X?

Barely, if the only thing you add is a question mark. I checked my own posts: the 438 that end in a question mark had a median of 57 replies and 18.1 replies per 1,000 impressions. The 792 that do not had 49 replies and 17.3 per 1,000. Across about 54,000 original posts from 669 ClimbX users (last 180 days, reach against each author’s own median), ending on a question was neutral for reach too (1.00x).

The type of question matters much more. In the outlier study, posts that asked readers to connect or answer a quick hook drew far more replies per view than general questions, and how-to playbooks drew almost none. Nearly all of the reply-heavy outliers came from accounts under 25,000 followers, which makes this one of the few areas where small accounts have the edge.

The pattern behind the chart: a reply is cheap when the reader already knows the answer and can give it in a few words. A connect call costs one line. A quick hook costs one word. A playbook asks the reader to evaluate your advice, which most people do silently with a bookmark.

7 post shapes people can answer in seconds

Each shape below lowers the cost of replying in a different way. Treat them as ideas to test, not a proven ranking: the samples are small, the shapes overlap, and they show what happened on one account. The weak and strong pairs are illustrative rewrites; the real posts link to the original on X.

1. The word-limit ask

Ask for an answer in one, two or three words. It is the strongest reply shape I have, on only six posts: they had a median of 21,102 impressions and 331 replies. In my label study of top versus bottom posts, word limits came with 7.25x my median reach and 6.35x my median replies. Six posts can be lifted by one lucky topic, so run your own handful before building a routine on it. The limit removes the fear of writing a wrong or long answer. My best example, “I am a founder scare me with 1 word”, drew 826 replies and 357 likes on 62,715 impressions.

2. The A-or-B pick

Offer two options and ask readers to choose. On my account it is a solid but not top shape: 28 posts, median 41 replies. Across ClimbX users under 5,000 followers it earned 1.10x their usual replies and 1.15x their usual likes (22 users). It works when both sides have real defenders; if one option is obviously right, nobody bothers to answer.

3. The audience-name question

Open with who the post is for: “Founders,” or “Indie hackers,”. My 29 posts that open this way had a median of 71 replies against 51 overall, and in the label study audience-name openers were four times more common in my top posts than in my bottom ones. Naming the reader tells them their answer is the one you want.

4. “Sell me X in two words”

Ask people to pitch their own thing in a tiny format. Everyone likes talking about their project, and the limit makes it a game. My “sell me your startup in two words max” post drew 669 replies and 304 likes on 51,998 impressions, and “pitch me your company in 1 word” drew 506. Posts that open with a direct ask like “tell me”, “name” or “pitch me” had a median of 101 replies across 37 posts, about twice my typical post.

5. The community call

“If you build in public, say hi” or “drop your project below”. My 27 community-call posts had a median of 169 replies, over three times my typical post, and 7,746 impressions. In the label study they had the highest win ratio of any type (4.1), and across ClimbX users under 5,000 followers connect posts earned 1.46x their usual replies (40 users).

Use it as a separate lane. Community calls give readers a stage, but they share no insight. Keep them to about one post in five at most, never two in a row, and expect the replies to be about the repliers, not about you.

6. The fill-in-the-blank

“The best advice I ignored was ___”. It looks like an easy reply, but on my account it underperformed: seven posts, median 31 replies. My guess from reading them is that the blanks were too open, so readers had to think too hard. It works better when the blank has an obvious shape, such as a tool, a number or a single word.

7. The specific-number question

Ask for a number the reader knows by heart: a price, a count, a date. One caution from my data: putting a number in the question did not help by itself. My 253 questions that contain a number had a median of 49 replies, against 58 for questions without one. The number should be the answer, not part of the setup.

More of my reply-heavy posts, with the numbers, are in my 10 biggest X posts. If you want to know which of these shapes also bring followers, see what to post on X to get followers.

Should you judge replies by count or by rate?

Use both, because they answer different questions. The count tells you how big the conversation got. The rate, replies per 1,000 impressions, tells you how willing the people who saw it were to answer, and it is the fairer way to compare posts with very different reach or accounts of different sizes.

Expect the rate to fall when a post spreads. My most-replied post, “USA has ChatGPT... Europe has?”, drew 6,439 replies from 1.36 million impressions: the biggest count I have, but 4.7 replies per 1,000 impressions, far under my median of 17.7. Once a post leaves your followers, it reaches people with less reason to answer. A high-rate post on 2,000 impressions tells you the shape works; a high-count post on a million tells you the topic travelled.

Neither number tells you whether the replies came from people you want to reach. Skim who answered: peers and potential customers in your niche are worth more than a pile of one-word replies from strangers.

What kills replies on X?

Anything that makes the reply cost more effort than it is worth. Five habits showed up clearly in my own posts.

• Vague closers like “Thoughts?” They ask for an opinion without saying on what. Replace them with a specific question, or let the post be the question.
• 26 words or more, on my account. My long posts had a median of 38 replies against 63 for 9 to 15 words, and in the label study 26+ word posts were half as common at the top as at the bottom. Many ClimbX users see the opposite for reach, so check your own numbers before cutting.
• “Hot take:” labels. Posts with the label had a median of 38 replies, and the label had a win ratio of 0.53. A real hot take does not need a sign announcing it.
• Links in the post. The 90 of my 1,230 posts with a link had a median of 24.5 replies and 13.5 per 1,000 impressions, the steepest drop of all. Put the link in a reply.
• Questions only you can answer. Expert and playbook posts had win ratios of 0.17 and 0.09 in my label study, and 0.21 replies per 1,000 views in the outlier study. Advice does get replies, from people who add a correction, ask a follow-up or share how it went for them, but that is a much smaller group than everyone who can name one word.

What should you do after the replies come in?

Reply to the first ones yourself, within the first hour. It shows readers the post is a real conversation, so the next person is more likely to join, and author replies were the kind of conversation X’s 2023 model valued most. Answer with something specific to their reply, not “great point”.

After a month, sort your own posts by replies and look at the top ten. Most people find two or three shapes doing the bulk of the work, and those become the daily default. If you want to see how replies fit with likes, reposts and bookmarks, the guide on how to get engagement on X covers each one separately. ClimbX’s daily drafts lean on these shapes: in a September 2026 review, question drafts were the type users posted most often (7 of 35, about one in five, a small sample).

How these numbers were measured

My posts: 1,230 original posts on @DanielSmidstrup from the last 180 days, queried October 3, 2026, with the latest metrics for each post (replies to other people excluded). Replies per 1,000 impressions is the median of each post’s own rate.

Label study: my top 500 posts against my bottom 200 by reach; a win ratio is a label’s share of the top group divided by its share of the bottom group.

Outlier study: 500 posts that beat their author’s normal reach, from 181 creators; reply rates shown for accounts under 25,000 followers.

ClimbX users: anonymized aggregates, my account excluded. The length and question figures use about 54,000 original posts from 669 users with 20 or more posts each, last 180 days, posts older than 3 days, lift = impressions divided by the author’s own median. The A-or-B and connect figures compare replies with each user’s own typical post, for users under 5,000 followers. No individual user’s post is shown.

Common questions about getting more replies on X