Bluesky Post Formats: What Your Analytics Can Tell You
There is no account-independent rule that image posts get twice the engagement, videos should run for 30 seconds, or three hashtags improve every Bluesky post. Those numbers usually come from a vendor's mixed sample, a different date range, or another network.
Use the formats Bluesky supports, label your own posts, and compare posts that were trying to do the same job. Your account data can tell you whether an image helps your release posts. It cannot prove that images are best for everyone on Bluesky.
The Bluesky post formats worth separating
Bluesky posts share one record type, but the format changes through embeds and relationships to other records. Keep these groups separate in an account-level review.
Text posts
A Bluesky post supports up to 300 grapheme clusters. A grapheme is closer to what a reader sees as one character, so some emoji sequences count differently from their raw byte or code-point length.
Text-only posts are useful for short observations, questions, updates, and conversation starters. Do not use them as the baseline for every other format without accounting for subject and purpose. A one-line question and a product announcement are not equivalent tests.
Image posts
Bluesky's official post tutorial allows up to four images in one post. Each image can include its own alt text and is limited to 2 MB at the protocol level documented by Bluesky.
Use the aspect ratio that fits the material. A square is not automatically better than a portrait image. Check the post on a narrow screen, keep important text away from the edges, and write alt text that describes the useful content rather than repeating the caption.
Strip image metadata before upload when location or device information should not travel with the file. Bluesky's developer tutorial recommends stripping metadata and notes that clients are currently responsible for sanitizing uploads.
Video posts
A post can contain one video embed. Bluesky processes video asynchronously, so an upload can finish before the video is ready to display. A scheduler needs to wait for processing and report a failure rather than publishing an empty post.
Bluesky-hosted accounts need a verified email address to upload video, and the video service applies additional account-level limits. Check Bluesky's current video upload documentation before building a workflow around a duration, file-size, or daily-upload number; those service limits can change separately from the post schema.
Add captions when speech or meaningful sound is part of the video. Inspect the first frame and the caption timing on mobile before scheduling it.
Link-card posts
A URL in the text is not automatically a rich preview in the protocol. A client builds an external embed with a URL, title, description, and optional thumbnail.
Check the final card before publishing. A stale Open Graph title or cropped thumbnail can make a good post look broken. Add a sentence that explains why the link matters instead of relying on the card to provide the whole post.
Quote posts and replies
A quote post embeds a reference to another record. A reply stores references to the parent and root posts in the thread. Both are different from an original standalone post because they begin with existing context and often reach a different audience.
Track replies separately from original posts. Fifty useful replies can be good community work, but mixing them into an original-post average makes the account's publishing pattern hard to read.
Hashtags and mentions
Hashtags and mentions are facets inside the post's rich text. Use a hashtag when people browse or search that topic. Do not add two or three because a generic checklist says that count performs best.
Mentions create notifications. Mention someone when the post is relevant to them. Borrowing an audience is not a useful reason.
Decide what each post is supposed to do
Choose the job before looking at the metrics. Common jobs include:
- start a useful conversation
- explain a product change
- send readers to an article or release note
- show a visual example
- answer a customer or community question
- report news or an event update
The useful metric follows from the job. Replies matter for a question. Link visits matter for an article. Reposts and quotes matter when the post is meant to travel through other networks. A release post may be useful with modest engagement if the right customers read the details.
Do not rank all posts by likes and call the first format the winner.
Build a small format comparison
Use a spreadsheet or a database table with one row per original post. Record:
| Field | What to record |
|---|---|
| Date and time | The account's local publishing time |
| Format | Text, image, video, link card, or quote |
| Job | Question, release, tutorial, news, promotion, or another specific label |
| Subject | The topic or campaign |
| Likes | Current like count at the chosen measurement time |
| Replies | Current reply count |
| Reposts and quotes | Keep both counts instead of merging them blindly |
| Link visits | Use a tagged URL when the post is meant to send traffic |
| Follower change | Account-level context, not credit assigned to one post |
Measure every post after the same interval, such as 48 hours or seven days. If one post is measured after an hour and another after a month, the comparison is not useful.
Keep the sample understandable. Ten release posts with and without images can teach you something about release posts on that account. Ten unrelated posts across six formats mostly tell you that the subjects were different.
Run one change at a time
Start with a question the account can answer:
Do screenshots help product-release posts earn more qualified replies and link visits?
Then run a simple comparison:
- Collect several recent release posts and label the format.
- Keep the subject, writing style, and call to action reasonably similar.
- Alternate text-only and screenshot posts instead of testing one format only on launch day.
- Publish across comparable audience windows.
- Measure after the same interval.
- Read the replies and inspect link visits, not only likes.
If the screenshot posts do better, keep testing them for release posts. Do not turn that finding into "images always win on Bluesky."
Keep the comparison fair
Separate timing from format
A morning image post and an evening text post test two variables at once. If the image performs better, you cannot tell whether the format or the audience window mattered.
Hold the posting window steady while comparing formats. Test timing in another round after you have enough comparable posts.
Ignore charts that name one universal best time without describing the accounts, time zones, date range, and metric. Bluesky offers multiple feeds with different selection rules, and an account's followers do not share one global schedule.
Account for outliers and small samples
One widely shared post can dominate an average. Keep the median beside the average, and open the best and worst posts before drawing a conclusion.
Small samples should change behavior cautiously. If two image posts outperform two text posts, run another comparison before making a permanent content rule.
Also note outside events. A product launch, election, outage, conference, or reply from a large account can explain more than the format.
Use account-level data and choose the next change
Use TheBlue.social for the account-level view
TheBlue.social's Bluesky analytics shows original posts and replies, likes, replies, reposts, quotes, top posts, follower history, and posting-time activity for a connected account. The page supports 7-, 30-, and 90-day views depending on access.
Use those metrics as the account-level record, then label post format and job in your comparison sheet. The dashboard does not make an unsupported platform-wide claim that one format wins.
The activity heatmap can help choose audience windows for a later timing test. It shows when the connected account's content received engagement; it does not guarantee future reach at a particular hour.
What to change after the review
Make one concrete adjustment:
- Keep text-only questions when they consistently start the right conversations.
- Add screenshots to release posts when readers need to see the change.
- Shorten or recut videos when viewers miss the useful part.
- Fix link metadata when the card is unclear or stale.
- Separate replies from original posts in future reports.
- Stop using a hashtag that brings unrelated replies or no useful discovery.
Run the next group with that change. Keep the labels and measurement interval the same so the next comparison is easier.
Frequently asked questions
Which Bluesky post format gets the most engagement?
There is no reliable universal winner. Compare formats within one account, for the same type of post, over the same measurement window. A format that works for an illustrator may not answer anything for a local reporter or software product.
Are image posts better than text posts?
Images help when the visual carries useful information: a product screen, chart, photograph, diagram, or piece of art. Adding a generic image to a text point can create more work without improving the post.
How many hashtags should I use on Bluesky?
Use the hashtags that lead to a relevant search or community. There is no fixed two- or three-hashtag rule that applies to every account and post.
Should replies count in post-format analytics?
Track them, but keep them separate from original posts. Replies begin inside another conversation and have a different distribution path from standalone text, image, video, or link posts.
How long should a format test run?
Run it until you have several comparable posts in each group and can inspect more than one unusual success or failure. For accounts that publish a few times per week, that usually means several weeks rather than a two-day test.
Last updated: August 12, 2026