Bluesky engagement rate
Engagement rate is a post's reactions set against the audience that could have seen it. Here is the formula Updrift uses, and why the number to beat is your own.
The formula
For one post: (likes + reposts + replies + quotes) ÷ followers × 1000, where every count is taken in the post's first 24 hours and followers is the number the account had on the day the post went out. The result is engagement per thousand followers, and it is the figure Updrift shows for every post.
A worked example. A post goes out when the account has 2,400 followers. In its first day it gets 31 likes, 6 reposts, 9 replies and 2 quotes: 48 reactions. 48 ÷ 2,400 × 1,000 = 20 per thousand. Another post, three months earlier at 800 followers, got 22 reactions: 22 ÷ 800 × 1,000 = 27.5 per thousand. The older post did better, though it got fewer likes.
Why the first day
Bluesky's Following feed is chronological. A post gets nearly all of the attention it will ever get in the hours after it goes out, and a post still collecting likes a week later is usually one that a repost or a feed picked up, which is worth knowing but is a different thing. Counting a fixed window does two things. It makes a post from last month comparable with one from this morning, because both are measured over the same span. And it rewards posting at the right hour, because a post at the wrong hour gets its engagement slowly if at all. Lifetime totals blur both.
Why per thousand followers
Raw counts grow with the account. Thirty likes is a quiet post for an account with 20,000 followers and a very good one for an account with 300. Dividing by followers on the day takes the size of the audience out of the figure, so a post from when the account was small can sit next to one from now, and a small account's figures read the same way as a big one's. Per thousand rather than as a percentage only keeps the numbers readable: 20 per thousand rather than 2%.
Followers on the day, not followers now. An account that has doubled since June would otherwise see every June post look twice as good as it was.
Why your own median beats any benchmark
The tempting question is what a good engagement rate is. Updrift does not publish one, on purpose. Across the accounts it measures, the spread between topics, sizes and countries is wider than the spread between a good post and a bad one on any single account, so a benchmark that averaged them would be a number nobody's posts were near. It would also drift with the network: Bluesky's feeds, its size and its habits have all moved in a year.
The comparison that holds up is with the same account's own posts. Take an account's recent posts, sort them by engagement per thousand, and the middle one is that account's usual. Every other post can be read as a percentage of it: 100% is usual, 180% is a post that did nearly twice as well, 40% is one that fell flat. That percentage is Updrift's lift figure, and it is the same kind of number for a knitting account with 500 followers and a news account with 50,000.
The median rather than the mean, because one post that travelled far drags the mean up until nothing seems to beat it. Half your posts beat the median by definition, which is what makes "beats your usual" a fair test.
What a typical spread looks like
Without naming a figure, the shape is consistent across the accounts Updrift measures. Most posts sit in a band around the median, a little above or a little below. A small number sit well above it, and they are the ones a repost, a feed or a busy reply thread carried. A few sit near zero: the ones that went out at the wrong hour or on a quiet day. The mean sits above the median because of the top few. When you look at your own figures, expect that shape, and read the posts in the top band for what they had in common: format, length, hour, tags, an image. That is what the What works report does by grouping them.
Working it out by hand
- Note the follower count each day, or at least on the days you post. Bluesky does not keep the history for you.
- A day after each post, note its likes, reposts, replies and quotes, and add them up.
- Divide by that day's followers and multiply by a thousand.
- After twenty or so posts, sort the figures and take the middle one. That is your usual. Read every post as a percentage of it.
Updrift does all four for one account free, reading the counts every hour so the first-day figure is exact, and shows each post's lift, the best hours and the hashtag table from them.
What the rate does not tell you
It does not measure reach, because Bluesky does not report it. It does not know why a post did well; a post carried by one big repost and a post that many people liked on their own get the same figure. And it says nothing about whether the people who reacted were the ones you wanted. It is a fair, simple figure for comparing your own posts with each other, and that is all it claims to be.
Questions
What is a good engagement rate on Bluesky?
One above your own median. Updrift has not published a benchmark figure, and any single number across accounts hides more than it shows: topic, size, country and time of day all move it. Compare your posts with your posts.
Does Bluesky show engagement rate?
No. It shows likes, reposts, replies and quotes per post, and followers on the profile. The rate is worked out from those, and it needs the follower count on the day the post went out.
Why not use impressions?
Bluesky does not report them, so no tool can. Followers on the day is the nearest honest denominator, and it is the same for everyone measuring the same account.
See your own figures, free.
Followers by the hour, the posts that landed, your best hours and how your hashtags do. One account, no card, for good.
The Planner, the Inbox and the full What works report come with a paid plan, 7 days free. See the plans.