A good engagement rate on X, measured by impressions, is 3% to 5%. Under 1% is low, 1% to 3% is where most accounts live, and above 5% is strong. Those are the bands the published studies agree on, and they are the ones X's own analytics screen should be read against.
Now the part that makes those numbers usable, because without it they are worse than useless. Engagement rate means two different things and people quote them interchangeably. One divides by impressions, one divides by followers, and for an account whose posts travel the second can be several times the first. A percentage with no denominator attached is not a measurement, it is a rumour.
The bands, by impressions
These apply to the by-impressions formula with a numerator of public actions: likes, replies, reposts and quotes. That combination is what the benchmark studies measured, so it is the only combination the bands can honestly be compared against.
Under 1% is the low band, and it usually points at the hook rather than the audience. Above 5% is worth interrogating rather than celebrating immediately, because a rate is a ratio and ratios improve for bad reasons too: a post shown to far fewer people often converts better among them, so a week where your reach collapsed can read as a week where your engagement improved. Always look at the rate next to the raw impression count, never alone.
The two formulas, spelled out
Both share a numerator. Only the denominator moves, and the denominator is the entire argument.
- Add up the public actions. Take likes, replies, reposts and quotes from the post and add them together. That total is the numerator of both formulas. For the example post used here, 168 likes plus 21 replies plus 34 reposts plus 5 quotes gives 228.
- Divide by impressions for the X analytics version. Divide that total by the impressions the post received. On 12,400 impressions, 228 divided by 12,400 is 0.0184.
- Multiply by 100. Turn the decimal into a percentage: 0.0184 times 100 is an engagement rate of 1.84% by impressions. This is the figure to compare against the published bands.
- Divide by followers for the media kit version. Take the same total of 228 and divide it by your follower count instead, then multiply by 100. On 4,200 followers that is 5.43% by followers, which is the number an influencer marketplace or sponsorship brief usually means.
- Say which formula you used. Quote the rate with its denominator attached, as in 1.84% by impressions or 5.43% by followers. The two differ by exactly the factor your reach exceeds your audience, so an unlabelled percentage is not a claim anyone can check.
One post, two honest answers, and a gap of nearly three times between them. That gap is information rather than noise: 12,400 impressions on 4,200 followers means the post was seen roughly three times over, so it travelled well past the people who already follow the account. By impressions answers "of the people who saw this, how many did something", which measures the post. By followers answers "how live is this audience", which measures the account, and that is why a brand buying access to your followers wants it expressed against the thing they are buying.
Why smaller accounts routinely score higher
Engagement rate falls as accounts grow, and it falls hard, which makes comparing across sizes close to meaningless. A small account is read mostly by people who chose it deliberately. A large one is read mostly by people the recommendation system put it in front of, and a reader who did not opt in is a reader far less likely to reply. The rate is measuring the mix of those two groups at least as much as it is measuring your writing.
The useful comparison is against yourself. Take your own last 28 days against the 28 before it, with the same formula and the same numerator both times. That comparison holds account size constant, which is the variable that otherwise dominates everything.
What actually counts as an engagement
The standard numerator is public actions only: likes, replies, reposts and quotes. X's own analytics counts more than that, including link clicks, profile visits, detail expands and hashtag clicks, so the engagement rate on your screen will be higher than the one you calculate by hand. Neither number is wrong, but comparing a benchmark measured one way against a rate computed the other way is, and it is the most common mistake people make with this metric.
Bookmarks sit outside the standard formula too. The bands were set before bookmark counts were public, so folding bookmarks in gives you a number that flatters you against everyone you are comparing yourself with. Track them alongside instead: a bookmark is a private signal that someone meant to come back, and a post with an unusual bookmark count is usually a post worth expanding into a full thread later.
Replies deserve more attention than likes
A like costs nothing and means roughly nothing. A reply costs the reader time, a sentence and a small amount of social risk, and early ranking prices those actions accordingly: replies weigh heaviest, then reposts, then likes, with profile visits and long dwell counting as well. Two accounts sitting at the same 2% are not in the same health if one of them is carrying ten times the reply share.
So read replies as a share of your total engagement, not just as a raw count, and treat it as the number you are actually trying to move. It also has the useful property of being directly actionable: posts that ask something answerable get replies, posts that conclude get likes. Choosing which of those to write more often is the practical half of deciding what to post.
What to do when the number is low
A low rate is almost never an audience problem, which is the conclusion most people jump to. It is one of two other things, and the raw numbers tell you which.
A hook problem looks like healthy impressions and thin engagement. The post was shown and scrolled past. The fix is in the first line and in whether the post gives anyone something to do: read back your best few posts and work out what they had that the rest did not. Length is often part of it, and a post that gets truncated mid-thought loses people at exactly the wrong moment, which is its own failure mode covered in why posts get cut off on X.
A reach problem looks like low impressions in the first place, and the rate may even be fine within that small sample. Then the question is what is throttling distribution: posting into hours where nobody is awake to produce the early signal, posting in bursts rather than on a rhythm, or leaning on formats the feed does not carry well. Whether posting time is part of it is worth fifteen minutes of checking before you rewrite anything.
Where to find these numbers in X analytics
All of it is available on a free account. For a single post, tap the small bar-chart icon underneath it. The panel that opens gives you impressions, total engagements and the breakdown into likes, replies, reposts, quotes, bookmarks and profile visits, which is every field either formula needs except your follower count.
For a period rather than one post, open Account overview from the top of that same panel and set the range to the last 28 days. Seven days is not enough posts to mean anything and ninety reaches back past whatever you last changed. Divide the totals by your post count and you also have engagements per post, which is the blunter number worth watching alongside the rate.
Then stop measuring and go back to writing, because the rate is a scoreboard and not a lever. The formats that move it, and the rest of the mechanics of running the channel, are at the X guides hub.
Frequently asked questions
What is a good engagement rate on X (Twitter)?
Measured by impressions, which is the formula X analytics reports, under 1% is low, 1% to 3% is typical, 3% to 5% is good, and above 5% is strong. Those bands assume a numerator of public actions: likes, replies, reposts and quotes. They are soft in one specific way, which is that rates fall as accounts grow, so a 2,000-follower account at 4% and a 200,000-follower account at 1% are not two different levels of skill.
How do you calculate engagement rate on Twitter?
Add likes, replies, reposts and quotes together, divide by impressions, then multiply by 100. A post with 168 likes, 21 replies, 34 reposts and 5 quotes on 12,400 impressions is (168 + 21 + 34 + 5) divided by 12,400, times 100, which is 1.84%. The other version of the formula divides that same total of 228 by your follower count instead, so on 4,200 followers it gives 5.43%. Both are called engagement rate, so always say which one you used.
What is the average engagement rate on X?
Most accounts sit between 1% and 3% by impressions, and that band is where the published studies put the average. Treat it as a range rather than a target. The same account will move around inside it week to week depending on format, subject and how far its posts travelled past its own followers.
Why is my engagement rate by followers so much higher than by impressions?
Because your posts are reaching past the people who follow you. The numerator is identical in both formulas and only the denominator changes, so the two rates diverge by exactly the factor your reach exceeds your audience. A post with 12,400 impressions on 4,200 followers was seen roughly three times over, which is why dividing by followers produces a much larger number. That is also the version media kits ask for.
Do bookmarks count toward engagement rate?
Not in the standard formula. The benchmark bands everyone quotes were set before bookmark counts were public, so adding bookmarks to your numerator gives you a number that flatters you against every account you are comparing yourself with. Track bookmarks separately instead, because a high bookmark count is a genuine signal that people meant to come back to the post.
Is a 1% engagement rate on X bad?
No, it is the bottom of the typical band rather than a failure, and for a large account it is close to expected. What matters more is the direction over a few months and the mix underneath it. A 1% rate carrying an unusual share of replies is healthier than a 2% rate that is almost entirely likes, because replies are the costliest action a reader can take and the one early ranking weighs most.