Almost every argument about engagement rate is really an argument about the denominator. One person is dividing by impressions, the other is dividing by followers, and the two numbers differ by around a hundred times. Both people think the other one is doing well or badly, and neither is measuring the same thing.
So before any benchmark is useful, you have to say which calculation you mean.
Two rates, two questions
Engagement rate per impression is engagements divided by the number of people who saw the post. It answers a question about the post: of the people it reached, how many did something about it? It is the number to use when you want to know whether a piece of writing worked.
Engagement rate per follower is engagements divided by your follower count. It answers a question about the account: of the people who chose to follow me, how many still respond when I say something? It is the number brands and sponsors usually mean, because it is a rough proxy for whether an audience is real.
Engagements themselves are likes, reposts, replies and bookmarks. The engagement rate calculator does both calculations and, importantly, tells you which one you are looking at.
Realistic numbers
Measured per impression, most posts land somewhere between 0.5% and 1.5%. Below 0.5% suggests the post reached people who kept scrolling. Above 4% is genuinely strong and worth studying, because something in it did work.
Measured per follower, the typical range is far lower, roughly 0.02% to 0.09% for an account posting regularly to a real audience. Anything above 0.3% usually means a small and unusually attentive following.
Treat those as orientation rather than targets. Ranges vary enormously by niche, by account size, and by how much of your follower list is still alive. A technical account with 2,000 engaged readers and a lifestyle account with 200,000 mostly dormant ones can produce identical engagement in absolute terms and wildly different rates.
Count bookmarks
Bookmarks are the most underrated signal on the platform and most calculators still leave them out, because bookmark counts were not publicly visible for years.
A bookmark is a stronger statement than a like. A like is a reaction that costs nothing; a bookmark is someone deciding your post is worth coming back to. It also feeds ranking. If what you write is useful rather than entertaining, bookmarks may be your single largest engagement category, and a calculation that omits them will make a genuinely valuable account look average.
Why the number falls as you grow
The most common panic is watching engagement rate decline month after month while everything else improves. This is almost always arithmetic rather than decay.
Follower counts effectively only go up. People who lose interest do not unfollow, they just stop opening the app, and their accounts sit in your denominator forever. Meanwhile every new follower is, on average, slightly less interested than the one before, because the earliest followers are the ones who sought you out.
So the denominator grows faster than genuine attention does, and the rate falls even as absolute engagement climbs. This is the single strongest argument for tracking engagement per impression instead. That number is not polluted by people who left in 2023.
What a low rate is telling you
The useful move is to read the two rates against each other, because the combination localises the problem.
A low rate per impression means the post reached people and did not move them. That is a writing problem, and specifically an opening problem: the first line did not earn the second. Posts fail in the first two seconds far more often than they fail in the conclusion.
A healthy rate per impression with a low rate per follower means the writing works but distribution does not. Your posts land well with whoever sees them, and not enough people see them. That is not fixed by writing better.
Both rates low usually means the account is talking into a niche it has not actually reached yet, which is a positioning problem rather than either of the above.
Fixing the distribution case
The distribution problem is the one most people have, and posting more rarely fixes it. Posting more shows more content to the same audience, which raises impressions slightly and lowers rate per impression, which feels like going backwards.
The thing that does work is replying. A reply on a post that is climbing puts your writing in front of an audience that is not yours, at the moment they are already paying attention to that topic. It is the only reliable way to get distribution from inside a small account, and it is why accounts that grow fast are almost always accounts that reply a lot.
The cost is that it does not scale by willpower. Finding conversations worth joining, early enough to matter, means watching the platform far more than is healthy. That is the specific problem ReplyBoss exists to solve: it monitors the accounts and keywords you care about, drafts replies in your voice, and leaves the judgement to you.
Measuring it honestly
Two habits make the number worth tracking.
Always state the denominator. "3.2% engagement" means nothing on its own and is frequently used, deliberately, to make per-impression numbers look like per-follower ones.
Compare like with like. A thread and a one-line joke are different products, and averaging them tells you nothing about either. Track them separately or the average will just hide whichever one is working.
When you want the number, the calculator runs in your browser, includes bookmarks, and shows the band your result falls into for the basis you picked rather than a single benchmark that ignores the question.