Bluesky analytics: what you can actually measure

Bluesky has no analytics dashboard, and the API exposes less than you would expect. An honest account of what is measurable and what is not.

Bluesky does not have an analytics dashboard. There is no equivalent of Instagram Insights or Page analytics, and third-party tools cannot invent data the network does not publish.

That constraint is not a gap waiting to be filled by a clever tool — it is a design consequence. Worth understanding before you pay anyone for Bluesky analytics.

What is measurable

• Likes, reposts and replies per post. Public, countable, reliable.
• Follower count over time, if something records it — Bluesky itself does not keep the history.
• Your own posting cadence, which sounds trivial and is the number most people are wrong about.
• Which posts travelled, by comparing engagement across your own posts.

What is not

Impressions and reach. This is the big one. Bluesky does not publish how many people saw a post, so nothing can report it. Any tool showing you Bluesky reach is estimating, and should say so.

Follower demographics. No age, location or interest data. It is not collected in a form anyone can query.

Link clicks, unless you measure them yourself with a redirect or UTM parameters at the other end.

Profile views. Not exposed.

Why the chronological feed makes this less painful

On an algorithmic network, impressions matter enormously because the algorithm decides your reach and you are trying to reverse-engineer it. On a chronological feed, your reach is approximately your follower count — the feed does not choose.

"You are not missing a number that would change your decisions. On a chronological feed, engagement per post and posting time are most of the signal."

What to actually track

• Engagement per post, not totals. Totals go up because you posted more, which tells you nothing.
• Time of day against engagement. The highest-value thing you can learn on a chronological network, and it needs several weeks of varied posting to see.
• Replies as a share of engagement. A high ratio means conversation rather than passive scrolling, and it predicts follower growth better than likes.
• Follower growth per week, recorded by something, because Bluesky will not remember it for you.

None of that requires a paid tool. It requires something recording the numbers over time, because the network does not.

How to record it without paying for anything

The whole problem is persistence, not access. Engagement counts are public on every post, so a spreadsheet updated weekly will get you most of the way: date, post, likes, reposts, replies, and your follower count that day.

That sounds primitive and it is genuinely enough for the first few months. The reason people move to a tool is not better numbers — it is that manually recording them survives about three weeks.

Reading the numbers without fooling yourself

Two traps are worth naming, because both are easy and both lead to worse posting.

Small samples look like signal. One post doing four times your usual numbers is far more likely to be one influential person reposting it than a repeatable insight about what your audience wants. Look at a month, not a post.

Follower count is a lagging, noisy number. It moves for reasons unconnected to what you posted — someone added you to a starter pack, an account you follow mentioned you, a spam wave got purged. Treat a week-over-week trend as information and a single day as noise.

What to do with what you learn

• If replies are low but likes are fine, you are broadcasting. Ask more questions; post fewer conclusions.
• If one time slot consistently outperforms, move more posts into it rather than adding new slots — though across 344 accounts we could not find a slot that reliably did.
• If engagement per post falls as volume rises, you are past your sustainable rate — see how often to post.
• If nothing correlates with anything, you probably do not have enough posts yet. Two months is roughly the point where patterns become readable.