How do you increase engagement on X?
Give people a reason to reply, make the post worth reading to the end, and keep the conversation going yourself. Four points in X's own documentation back that up (checked 4 November 2026):
- Answer the people who reply. X's conversations help page lists factors it considers when ranking a reply higher, starting with “if the original post author has replied”, then whether the reply comes from someone the viewer follows and whether the replier subscribes to Premium. Answering is the one you control.
- Write for time spent. In updates dated 8 and 13 October 2026, X's algorithm README says its ranking model now does a better job of estimating how long you'll spend on each post, and that retrieval, which finds posts from accounts you don't follow, is also trained to surface posts people read all the way through.
- Aim past the like. The ranking model predicts replies, reposts, quotes, shares by DM or copied link, profile clicks, dwell time and following the author as well as likes, plus negatives such as not interested, mute, block and report. How those predictions are weighted is in our post on X's algorithm and the mutual-follow boost.
- Don't trade likes for blocks. The README describes offline jobs, called agatha, that label an account from how others respond to its posts, using blocks, reports and spam reports relative to likes. Those labels are among the inputs to visibility filtering, so bait that provokes blocks and reports can work against you even when the like count looks good.
What do impressions, views and engagements mean on X?
Impressions count how many times a post appeared on a screen, views count looks at a post by logged-in people, and engagements are a private total X gives only for your own posts. None of them counts unique people. From X's API metrics page and view counts help page, checked 4 November 2026:
| Metric | What X says it counts | Watch for |
|---|---|---|
| Impressions | Times the post appeared on a user's screen. Not unique: the same person seeing it twice counts as two. | The usual base for engagement rate. |
| Views | Anyone logged in who views the post, wherever they see it and whether or not they follow you. Your own looks count too. | Not all unique: a look on the web and another on your phone count as two. Embedded posts don't add views, and Community posts, X Circle posts and older posts show no view count. |
| Engagements | A private total for posts you own, listed beside URL clicks and profile clicks. | The API returns private metrics only for posts created in the last 30 days, and the page doesn't list which actions make up the total. |
| Likes, replies, reposts, quotes, bookmarks | Public counts anyone can see. The repost count leaves out quote posts. | Quote posts are counted on their own, and the page says they are always organic. |
| Video playback at 0%, 25%, 50%, 75% and 100% | Unique users who played through each point. | X points to them for understanding drop-off. |
So are X's views accurate? They count what X says they count, repeat looks and your own included. Read views as reach, and judge a post by engagements per impression.
Is a 12% engagement rate good on X?
It can be, but the figure means little until you know the base and your own usual rate. Engagement rate is engagements divided by a base, and our engagement rate glossary entry lists the common ones: impressions, reach or followers. Take two posts from a sample account, Northwind Ltd, with 1,500 followers (invented numbers):
| Sample numbers | Post A | Post B |
|---|---|---|
| Impressions | 400 | 6,000 |
| Engagements | 48 | 240 |
| Rate by impressions | 48 ÷ 400 = 12% | 240 ÷ 6,000 = 4% |
| Rate by followers | 48 ÷ 1,500 = 3.2% | 240 ÷ 1,500 = 16% |
By impressions, Post A wins at 12%. Yet Post B appeared on 15 times as many screens (6,000 ÷ 400) and earned 5 times the engagements (240 ÷ 48), and by followers the order flips. A 12% rate on 400 impressions can simply mean the post never left the people who already know you.
The X help pages and API documentation we checked give no benchmark rate (checked 4 November 2026), so set your own. List your last 20 posts with impressions and engagements, divide each pair, and take the middle value, the median. A new post well above your median is a topic or format to repeat; someone else's 12% tells you little about your account.
Which post formats get the most engagement on X?
No single format wins; pick it by the response you want. The deciding facts come from X's help pages, checked 4 November 2026:
| Format | Use it when | What X's help pages say | Watch out for |
|---|---|---|---|
| Poll | You want a quick read on a choice your buyers face | Up to four options of 25 characters each; runs from 5 minutes to 7 days, 1 day by default; no photos (polls page) | Neither you nor other voters can see who voted or how, so a poll can't show which buyers answered. Ask for the reason in the replies. |
| Thread | The idea needs steps, such as a teardown or a how-to | In a follower's timeline, a thread of 2 or 3 posts generally shows as a connected bundle; at 4 or more, posts are truncated behind Show this thread (threads page) | Put the point and the number in the first post. |
| Single image | A chart, screenshot or before-and-after makes the point | On iOS and Android, a single photo with an aspect ratio between 2:1 and 3:4 displays in full, and images don't count toward the character limit (photos page) | Crop to that range so the whole image shows. |
| Hashtags | A real community follows a tag, such as an event's | X recommends no more than 2 per post, and anyone searching a hashtag may find a public post that uses it (hashtags page) | Excessive, unrelated hashtags count as content spam. |
| Limited replies | A post draws spam, or you want answers from specific people | Choose Everyone, Accounts you follow, Only accounts you mention or Only verified accounts, before or after posting (conversations page) | Fewer people can reply, and quote posts don't inherit the limit. |
| Quote post | You want all your followers to see your take on someone's post | The conversations page says only relevant people, such as those who follow you and someone in the conversation, see a reply in their Home timeline; to reach all your followers it suggests a repost or quote post | It reaches people who haven't seen the original, so it must make sense on its own. |
Posting links? Our post on whether X penalises links covers what X said, what its code shows and what OpenTweet's data on 28,124 posts showed.
Which engagement tactics break X's rules?
Trading or buying engagement, automated likes, and a few editing, hashtag and link tricks. From X's authenticity policy (April 2025) and automation rules (updated April 2026), checked 4 November 2026:
- Engagement swaps and pods. Coordinating to exchange engagement in any X feature, such as likes, polls, replies, reposts, Lists or views, is engagement spam, and so is paying others to inflate those numbers.
- Automated likes. The automation rules say you may not like posts in an automated manner, and warn that scripting the X website may result in permanent suspension.
- Bait-and-switch edits. Editing a post to use its existing engagement for substantially different content is content spam. X's own example turns a pancakes-or-waffles question into a request to like the post and subscribe for investing tips.
- Hashtag hijacking. Using trending or popular hashtags to manipulate a conversation or drive attention to your product is content spam, and so is posting excessive, unrelated hashtags.
- Bare links. Repeatedly posting links without commentary is content spam when that makes up the bulk of your activity.
- Asking for engagement, inside Original Content Rewards. If you're in X's Original Content Rewards programme, its rules bar repeatedly instructing people to like, reply, bookmark, follow or repost.
One cost is reach. Restricting reach is among the policy's enforcement options, and it can mean posts excluded from search and trends, removed from the For You and Following timelines, or downranked in replies. The rules on follower tactics are in our companion post on getting followers on X without buying them.
Why did my engagement on X suddenly drop?
Rule out the ordinary causes before assuming a penalty. Five checks, from X's help pages and code:
- Replies compared with posts. The README says For You filters out replies and reposts from accounts the viewer doesn't follow, so a week spent replying can show fewer impressions than a week of original posts.
- Community posts. X shows no view count on Community posts or X Circle posts, so a blank view count there is expected.
- Post age. For You filters out posts older than 48 hours, so a post's reach in For You ends after two days.
- Labels. X's myths page says “we don’t shadow ban! Ever.” Its authenticity policy still lists restricting reach as an enforcement option. The README describes a pilot tool, Under the Hood, that shows aggregate statistics on the visibility-impacting labels on your account and posts; X says it is expanding availability over time, so you may not have it yet.
- Blocks and reports. Mutes, blocks and reports are negative actions in the ranking model, and blocks and reports relative to likes feed the agatha account labels described above. Look at which recent post drew them.
What should a small B2B team do each week on X?
Thirty minutes a day and a review on Friday. The routine is our suggestion; the reasons come from the sources above.
Set your baseline
List your last 20 posts with impressions and engagements, work out each rate and note the median. X's API returns private metrics, engagements included, only for posts created in the last 30 days, so if you pull numbers through the API or a tool built on it, log them every week.
Post one thing worth reading to the end
One original post a day from your own work, with a number and a specific question your buyers can answer. Use the templates below.
Answer every reply the same day
The original author replying is the first factor X lists for ranking a reply higher. Add something in each answer: a detail, a number or a follow-up question.
Reply to five buyer posts
Spend the rest of the time under posts by people who match your buyer. Our guide to whether the reply-guy strategy still works has four reply shapes to adapt.
Review on Friday
Log five numbers: posts published, median engagement rate against your baseline, replies from likely buyers, profile clicks, and conversations moved to DM. Keep the topic and format that beat your median and drop the one that didn't.
For more, use our 32 X post and reply templates for founders.
Frequently asked questions
How much does X pay for 1 million views?
Neither of the two X help pages on creator pay gives a rate per view or per 1,000 impressions (checked 4 November 2026). Creator Revenue Sharing ended on 7 September 2026. Its replacement, Original Content Rewards, pays every two weeks on qualified impressions: unique impressions from Premium subscribers in the Home timeline with at least half the post visible. Requirements include Premium, 500,000 Home timeline impressions from verified users in 90 days, excluding replies, and 500 verified followers.
Do your own views count on X?
Yes. X's view counts page says that if you're the author, looking at your own post counts as a view. Views aren't all unique either: opening a post on the web and then on your phone counts twice. Only logged-in viewers count, and views of your post embedded on a website don't add to the total.
Does the reply count on X show every reply?
No. The number next to the reply icon counts direct replies to the original post, not every reply in the conversation, according to X's conversations help page. X also groups replies into sub-conversations and doesn't always show them in time order, so a busy thread can hold more replies than the counter shows.
How do you stop spam replies on an X post?
Limit who can reply. Before posting, tap Everyone can reply next to the globe icon and pick Accounts you follow, Only accounts you mention or Only verified accounts. After posting, open the More icon on the post and choose Change who can reply. You can also hide a reply: X says hidden replies move out of the main conversation but stay viewable through an icon on the post.
Why do some posts look bigger on my X profile?
Because they got more engagement. X's profile help page says posts that have received more engagement appear slightly larger, so your best content is easy to find. That makes your own profile a quick read on which topics landed: scroll it once a month and note which larger posts shared a subject or format.
Sources
- X Help Center: About conversations on X (checked 4 Nov 2026)
- X algorithm repository: README (For You feed) (checked 4 Nov 2026)
- X Developer Platform: Metrics (X API v2) (checked 4 Nov 2026)
- X Help Center: View counts (checked 4 Nov 2026)
- X Help Center: About X Polls (checked 4 Nov 2026)
- X Help Center: How to create a thread on X (checked 4 Nov 2026)
- X Help Center: How to post photos or GIFs on X (checked 4 Nov 2026)
- X Help Center: How to use hashtags (checked 4 Nov 2026)
- X Help Center: Authenticity policy (April 2025) (checked 4 Nov 2026)
- X Help Center: Automation rules (updated April 2026) (checked 4 Nov 2026)
- X Help Center: Original Content Rewards (checked 4 Nov 2026)
- X Help Center: Creator Revenue Sharing (checked 4 Nov 2026)
- X Help Center: Debunking X myths (checked 4 Nov 2026)
- X Help Center: How to customize your profile (checked 4 Nov 2026)