X gives every account a page of numbers and almost no guidance on what they mean. Impressions go up and followers do not. Engagement rate drops the week a post does well. A chart dips and you change your whole approach. Most of that confusion comes from reading the numbers as scores instead of as answers to specific questions.
This guide defines each metric in plain language, says which question it answers and which it does not, and gives a ten-minute weekly routine that uses four of them and ignores the rest. The goal is not more data. It is to stop changing things that work because a number moved.
What each number means
- Impressions: how many times a post appeared on someone's screen. One person scrolling past it twice is two impressions. It measures distribution, not interest.
- Engagements: every interaction with the post added together: likes, replies, reposts, quotes, bookmarks, link clicks, profile clicks, media opens, even expanding the post. A broad bucket that mixes very different behaviours.
- Engagement rate: engagements divided by impressions. It says what share of the people who saw the post did anything at all. Small accounts often see higher rates than large ones, because a small audience is mostly people who actually know you.
- Replies: someone wrote back. The most expensive action for the reader and the one that extends a post's reach, because the conversation keeps the post alive.
- Likes: the cheapest action. Pleasant, weakly informative. A post can get many likes and start no conversation.
- Reposts and quotes: someone put the post in front of their followers. Quotes with commentary are the strongest sign a post said something.
- Bookmarks: someone wants to come back to it. The clearest signal that a post was useful rather than merely agreeable.
- Profile visits: someone clicked through to your profile. The step between seeing you and following you.
- Follows: the outcome everyone watches, and the slowest to move.
Which question each metric answers
| Metric | What it tells you | What it does not tell you |
|---|---|---|
| Impressions | How far X distributed the post | Whether anyone cared. A post can be shown widely and ignored |
| Engagement rate | How interested the people who saw it were | How many saw it. A tiny post can have a high rate |
| Replies | Whether the post started a conversation | Whether it was good. Arguments get replies too |
| Bookmarks | Whether it was useful enough to keep | Whether it reached anyone new |
| Profile visits | Whether a post or reply made people curious about you | Whether they stayed. Visits without follows point at the profile |
| Follows per week | Whether the whole system is working | Which part is working. Cross it with the others |
Read the table as pairs. Impressions up and engagement rate down usually means a post escaped your usual audience into people who do not know you, which is normal and fine. Profile visits up and follows flat means the click is happening and the profile is not converting; fix the bio and the pinned post, not the posts. Each metric on its own is a half-answer.
Compare your posts with each other, not with benchmarks
Searching for a "good engagement rate on X" produces numbers from accounts that are nothing like yours: different size, different subject, different mix of followers. The only fair comparison for your post is your other posts. Sort your last thirty by replies and look at the top five. What shape were they? A question, a number, a decision, an image? Then look at the bottom five. The pattern is usually visible in minutes and it is specific to you.
The engagement rate calculator is useful for one thing in this process: working out a single post's rate quickly so you can rank posts by it. Use it to compare your own posts, then ignore the benchmark ranges once you have your own.
Sorted my last 30 posts by replies. Top 5: all questions about a decision I was actually making. Bottom 5: all advice. I have been writing the bottom 5 on purpose for a year.
The kind of finding this exercise produces. It is not a benchmark; it is a fact about one account that changes what that account posts next.
A ten-minute weekly routine
- 01Same day each week, open your analytics for the last seven days. Ten minutes, a timer helps.
- 02Replies per post: which two posts got the most? Write down what they had in common in one line.
- 03Profile visits and follows for the week: did visits lead to follows at roughly the usual ratio? If visits rose and follows did not, the profile is the problem this week.
- 04Best hours: which posts did well and when did they go out? After a few weeks a pattern appears. How to find your best time to post explains how much data you need before trusting it.
- 05Decide one change for next week. One. Then close the tab.
Numbers to ignore most of the time
- Daily follower count. It moves on noise: a big account's follower purge, a post of yours shown to the wrong crowd. Look weekly.
- Total impressions for the month. One post that travelled far dominates it and hides everything else.
- Likes as a quality signal. Likes reward agreeable posts. Replies and bookmarks reward useful ones. If you must choose, watch the second pair.
- Video views. X counts a view early in playback. The number looks large and tells you very little.
- Any metric on the day of posting. Posts keep collecting replies and bookmarks for a day or two. Judge after 48 hours.
How to read a dip
Every account has a bad week. The instinct is to find the cause in what you did, and sometimes it is there: you posted less, you posted at odd hours, you dropped the replies. Just as often it is not. The accounts you reply to had a quiet week, a holiday emptied the feed, or a post of yours two weeks ago reached strangers who inflated last week's baseline. Before changing anything, check three things.
- 01Did your activity change? Count posts and replies for the dip week against the week before. If they fell, the dip explains itself.
- 02Did the previous week have an outlier? One post that travelled makes the next week look like a fall when it is a return to normal.
- 03Is the dip in impressions only, or in replies and follows too? Impressions alone falling, with replies steady, usually means the feed was quieter, not that you were worse.
If your activity was steady and the dip lasts three weeks or more across every metric, then look at what changed in the posts themselves, and check the shadowban test to rule out a visibility problem. One bad week is weather. Three is a pattern.
From numbers to the next post
Analytics are only useful if they change what you write. The weekly routine ends with one decision, and the decision should usually be about shape, hour or frequency. If the top posts were questions, write two more questions. If the best hour is early evening, move tomorrow's post there. If a second daily post never gets replies, drop it and spend the time on replies to other people, as how often to post on X describes.
And when posts are being seen but not answered, the problem is usually in the post, not the numbers; how to get more engagement on X walks through the common causes. For where analytics sit in the whole picture, see how to grow on X.
BuzzyX · X growth workspace
Grow on X with posts that sound like you.
Compose in your voice, schedule the week, find posts worth replying to and read real analytics from the X API. $29/mo after the trial, everything included.
Frequently
asked questions
Something else on your mind? Get in touch.
What is a good engagement rate on X?
There is no single number that applies across accounts. Small accounts often see higher rates than large ones because their followers know them personally, and rates differ by subject and by post type. Compare your own posts with each other and look at what your top ones share; that is a better guide than any published benchmark.
Why did my impressions go up but my followers did not?
A post reached people outside your usual audience, who looked and moved on. That is normal. If it keeps happening, check whether those people are clicking through to your profile and leaving; if so, the bio and pinned post are not telling them why to stay.
Are likes or replies more important?
Replies. A reply costs the reader effort, extends how long a post is shown, and tells you what people actually think. Likes are pleasant and cheap. If you track only one engagement number, track replies per post.
How often should I check my analytics?
Once a week, for about ten minutes, on the same day. Checking daily makes you react to noise. Checking monthly means you miss things like a best hour you could be using now.
Do I need a paid tool to understand my X analytics?
No. X's own analytics page has every number in this guide. Tools help by keeping history, charting trends and working out best times for you, which saves the manual sorting, but the reading itself needs nothing but the routine above.
What do bookmarks tell me that likes do not?
That the post was worth coming back to. People bookmark things they intend to use: a method, a list, a reference. A post with more bookmarks than likes is usually your most useful one, even if it looked quiet.