Strategy
A Personal Brand Playbook for Each Social Network
What X, Threads, LinkedIn, Bluesky and Mastodon say about how they rank posts, what each one penalises, and what that means for what you post and how often.
Oleh Husiev12 min read
Pick a narrow topic first, then play each network by its own rules. They don't reward the same things. X's open-source ranker weighs a predicted reply about 7.5 times a like. Threads says replies drive almost half its views. LinkedIn is filtering engagement bait. Bluesky's main feed is chronological, and Mastodon doesn't rank posts at all.
Most personal branding advice treats "the algorithm" as one thing. It isn't. Below is what each of the five networks has actually published about how it ranks posts and what it penalises, with what that means for what you post. Where a network hasn't said something, we say so.
Start with what you're known for
Every network works better when people can tell in one line what you post about. Justin Welsh, who has written openly about building a solo business to $10 million in revenue, started with a very narrow audience: early-stage SMB SaaS founders in healthcare. LinkedIn has said something similar from its side: in a 2023 interview, its product team said writing within your core subject helps the feed route your posts to the people who want them.
If you haven't written that line yet, our guide to a personal brand thesis walks through it. The rest of this post assumes you know what you want to be known for.
X
What it ranks on. X open-sourced its current ranker, a Grok-based model, in January 2026, and keeps the code public. It predicts how likely each viewer is to take each action on your post and multiplies those probabilities by weights. The production defaults include:
| Action | Weight |
|---|---|
| Like | 0.66 |
| Reply | 5.0 |
| Quote | 5.0 |
| Share by DM | 5.0 |
| Share by copied link | 20.0 |
| Follow the author | 4.0 |
| Open a link | 0.2 |
| Not interested | -52.3 |
| Mute the author | -58.8 |
| Report | -234.0 |
Replies get an extra 15 on top when you and the reader follow each other. X's own code comments warn against reading these as exchange rates between likes and reports: the weights multiply predicted probabilities, not raw counts, and a report gets a large weight because reports are more than a thousand times rarer than likes.
What else the code does. Each post after your first one in someone's feed gets marked down, posts from accounts the reader doesn't follow are discounted, new accounts with few impressions get lifted, and posts older than 48 hours drop out of For You.
What it penalises. X's automation rules ban "duplicative or substantially similar posts" on one account or across accounts you run, automated likes, and automated replies triggered by keywords. Its spam rules cluster duplicate text in posts and replies.
What that means for you.
- Write posts people want to answer or send to a friend. A question you actually want answered, a specific claim people can push back on, a number with context.
- Spread posts out rather than posting five in a row; each extra one in the same feed counts for less.
- Links aren't penalised in the code, and X's leadership has denied deboosting them. Link posts still tend to do worse because they invite fewer replies, so say the point in the post and put the link second.
- The well-known "an author's reply is worth 75 likes" figure is from Twitter's 2023 code. The 2026 ranker doesn't predict that action at all.
- Free accounts get 280 characters; X Premium allows up to 25,000.
Threads
What it ranks on. Meta's system card for the Threads feed lists the predictions it uses: whether you'll like a post, reply to it, click the author's profile, how long you'll spend viewing it, and how deep the conversation under it goes, down to replies to replies. The Following tab is plain reverse-chronological.
What Meta says works. In its own guidance for creators, Meta says "replies account for almost half of the views on Threads", recommends "posting at least 2-5 times per week", and says posts with a topic tag "generally receive more views". Adam Mosseri, who runs Threads, put it more bluntly in a 2024 interview: "you should reply much more than you post."
What it penalises. Meta's card doesn't list penalties. Mosseri has said Meta is working to get engagement bait "under control", and that Threads doesn't downrank links on purpose but doesn't place much value on them, because people rarely like or reply to them.
What that means for you.
- Plan time for replies, on your own posts and other people's. That's where Meta says half the views come from.
- Add one topic tag. Threads allows exactly one per post.
- Two to five posts a week is Meta's own floor, and the only official posting number any of these five networks has published.
- Posts are 500 characters, and a text attachment can hold up to 10,000 more when you need room.
- Mosseri's own recurring "ask me anything" posts are a good model of a format built on replies.
What it ranks on. LinkedIn's feed team told Entrepreneur in 2023 that it prioritises posts "grounded in knowledge and advice", shows people more posts from those they follow, and treats sudden virality as a reason to look closer. Its engineering team has written about measuring dwell time, the time a post is at least half on screen, and about a model that marks down posts people are likely to scroll past. In 2026 it moved most of the feed to a new transformer-based ranker that, in its tests, increased time spent on the feed.
What it penalises. In March 2026, LinkedIn's VP of Engineering wrote that it's "working to make engagement pods ineffective and curb comment automation" and will "filter out engagement bait", with "Comment 'Yes' if you agree" as the example. Its Help Center says it may limit how many comments an account makes and limit the visibility of automated comments.
What that means for you.
- Teach something specific from your work. A post someone saves because it's useful is what LinkedIn says it wants.
- Write so people keep reading: a clear first line, short paragraphs, one idea.
- Comment like a person on other people's posts, in your own words. Pods and comment tools are exactly what LinkedIn is building against.
- On frequency, Buffer's analysis of over 2 million LinkedIn posts found that accounts got more impressions per post in weeks they posted more, compared with their own one-post weeks. That's correlation in one tool's users, but it's the best data available.
Our LinkedIn strategy guide goes deeper on profiles and post types.
Bluesky
How the feeds work. Bluesky's default Following feed is chronological. New accounts also get Discover, and anyone can build and publish a custom feed, which people pin like tabs. Starter packs let anyone recommend up to 150 accounts and three feeds to new users in one click.
What it penalises. Bluesky's community guidelines ban spam and artificially manipulating engagement. Since October 2025 it also down-ranks replies it detects as toxic, spammy or off-topic, and uses a private "dislike" signal to personalise Discover.
What that means for you.
- Get into the feeds and starter packs for your field. On a chronological network, being in the right custom feed is how strangers find you.
- Reply with substance. Low-effort replies are now down-ranked rather than rewarded.
- Posts are 300 graphemes, so write tight or split into a thread. Bluesky's COO has been quoted saying it doesn't de-promote links, though how Discover ranks posts isn't public.
Mastodon
How discovery works. Mastodon's own site promises "no algorithms or ads". Your home feed is chronological, posts reach new people through boosts and hashtags, and since version 4.2 people can opt in to having their posts appear in search and Explore. Full-text search is optional for each server, and Mastodon 5.0 (August 2026) is de-emphasising followed hashtags ahead of folding them into custom feeds.
What it penalises. There's no ranking to penalise you, but every server has rules. A few we checked through their own rules pages on October 20, 2026:
- fosstodon.org: don't use automated tools to post "without also monitoring and/or interacting from your account", and no posts that are only links or excessive hashtags.
- mastodon.social: generative AI use must be disclosed, and accounts may not post only AI-generated content.
- infosec.exchange: mostly automated accounts must be marked as bots, and automated public posts are capped at one an hour.
What that means for you.
- Use a few hashtags in CamelCase (#BuildInPublic, not #buildinpublic) so screen readers can read them.
- Opt in to search and Explore in your profile settings if you want strangers to find you.
- Answer replies. A scheduled post from a person who replies is normal on Mastodon; an account that only broadcasts is the pattern server rules are written against.
- Respect your server's content warning norms and add alt text to every image.
Cross-posting without looking like a bot
None of these networks has a rule against posting the same idea elsewhere. X's duplicate rule covers your accounts on X, not your LinkedIn. What makes a copy look automated is the details:
| X | Threads | Bluesky | Mastodon | ||
|---|---|---|---|---|---|
| Length | 280, or 25,000 with Premium | 500, plus a 10,000 attachment | 3,000 | 300 graphemes | 500 by default, set per server |
| Mentions | @name | @name | Tagged names | @name.domain | @name@server |
| Hashtags | Optional | One topic per post | Optional | Inline or as tags | Main way to be found |
| Replies | Answer on X | Answer on Threads | Answer on LinkedIn | Answer on Bluesky | Answer on Mastodon |
The writer Molly White, who publishes on her own site and syndicates to Twitter, Mastodon and Bluesky, describes the approach that holds up: reformat for each network, split or link back where a limit is tight, and answer replies on each platform separately. Our cross-posting guide walks through it step by step.
Building in public, with the risks
Sharing numbers and decisions as you go still works as a format. Buffer has published its revenue and every salary since 2013, and keeps its metrics live at buffer.com/metrics. It has also written about the costs: transparency "can create rigidity" and "does inherently create extra work", and one salary mistake cost it some trust.
The risk has grown. Arvid Kahl, who shared his startup's revenue publicly before selling it, wrote in April 2026 that sharing revenue and architecture is now "outright dangerous" because coding agents make products cheap to clone. His test is a good one for any post about your work: "Interesting to participate in. Not easy to clone." Share the decisions, the lessons and the process. Think harder about the exact recipe.
Our building-in-public playbook has a weekly rotation and examples for each network.
How often to post
Nobody has shown that posting more makes you grow faster on any of these networks. What exists:
- Threads: Meta recommends at least 2 to 5 posts a week.
- LinkedIn: Buffer's data links more posts a week to more impressions per post, within the same accounts.
- Research: a 2026 study of 460 Instagram creators linked posting consistency to engagement, and earlier work by the same authors found both too little and too much posting went with lower engagement. Both are correlational, and both are about Instagram.
The practical answer is a pace you can keep for six months without hating it, with time set aside for replies. We cover setting that up in a publishing system that works on your worst days, and the per-network numbers in how often to post on social media.
Where Shaflex fits
Shaflex covers the mechanical parts of this playbook. The Composer gives each network its own copy of your draft with its own limit, so the X version, the Bluesky version and the LinkedIn version can differ. Split turns a long draft into a thread on X, Threads, Bluesky and Mastodon. Mastodon posts go out with their content warnings and alt text. And on Creator and up, replies from all five networks land in one inbox, so answering them takes one tab instead of five.
Sources
Checked on October 20, 2026.
- X: x-algorithm repository, production ranking weights, mutual-follow reply boost change, duplicate-text spam rules
- Twitter (2023): heavy ranker README
- X Help Center: Automation rules
- Nieman Lab: X makes overtures to journalists with a feature for links, October 2025
- Meta Transparency Center: Threads feed ranking system card
- Meta: Find your community with new Threads educational insights and new Threads features
- Instagram Help: Topics on Threads
- Meta: Attach text to your Threads posts
- Platformer: Adam Mosseri interview, July 2024
- Engadget: Threads users still barely click links, July 2025, and Meta on engagement bait
- Entrepreneur: LinkedIn changed its algorithm, June 2023
- LinkedIn Engineering: Understanding feed dwell time
- LinkedIn: Feed SR, the 2026 ranking model
- Tim Jurka, LinkedIn: Updates to the LinkedIn feed, March 2026
- LinkedIn Help: Comment limits and automated comments
- Buffer: How often to post on LinkedIn
- Bluesky: Viewing feeds, custom feeds, starter packs, October 2025 update, Community Guidelines
- Mastodon: joinmastodon.org, how the network works, posting, Mastodon 4.2, quick start guide on CamelCase hashtags
- Server rules, from each server's
/api/v1/instance/rules: fosstodon.org, mastodon.social, infosec.exchange - Molly White: POSSE
- Justin Welsh: My $10M journey
- Buffer: Open metrics and our salary system
- Arvid Kahl: The increasing risk of building in public, April 2026
- Tafesse and Hussain (2026): Steady posts, stronger bonds?, Journal of Marketing Communications
Questions readers ask
Which social network is best for a personal brand?
The one your audience already reads. LinkedIn suits professional and B2B audiences, X still suits tech, finance and media, Threads reaches a broad mainstream audience, Bluesky has strong developer, science and journalism communities, and Mastodon is strongest in open source, academia and privacy-minded circles. Start with one or two and add more once posting there is a habit.
How often should you post to grow a personal brand?
Often enough to keep a steady pace for months. The only official number from any of the five networks is Threads': at least 2 to 5 posts a week. Buffer's data on LinkedIn found more posts a week went with more impressions per post, while research on Instagram creators found both too little and too much posting went with lower engagement. Nobody has shown a causal link between frequency and follower growth.
Does posting the same text on every network hurt you?
Not because of a cross-network duplicate rule. X's rules against duplicate posts cover your own accounts on X, and X can't see your LinkedIn. The real costs are practical: mentions that don't resolve on other networks, the wrong length, hashtags that break one network's norms, and replies you never answer. Adapt each copy and answer replies where they happen.
Do links hurt reach on X, Threads or LinkedIn?
X's published ranking code has no link penalty, and opening a link carries a small positive weight. Threads' head, Adam Mosseri, has said Threads doesn't intentionally downrank links but doesn't place much value on them because people rarely reply to them. LinkedIn hasn't said either way. A link post usually does worse because it gives people less reason to stay and reply, not because of a rule.
Is building in public still worth it?
For many people, yes, but it's riskier than it was. Arvid Kahl, who sold his startup after sharing its revenue publicly, wrote in 2026 that sharing revenue and architecture is now dangerous because AI tools make products cheap to clone. His test for what to share: interesting to follow, not easy to copy.
