Thread splitter

Why threads still work

Threads persist because they solve a real problem: X's timeline rewards short posts, but ideas worth following someone for rarely fit in 280 characters. A thread lets you publish essay-depth material in a format the algorithm and the reader both tolerate. Mechanically, threads also compound engagement — each post in the chain is its own like-and-repost surface, and replies to any part push the whole thread back into circulation. A good thread is less a long post than a sequence of small, individually quotable posts that happen to build on each other.

Where to split: the craft part

The splitter above handles the mechanics — sentence-boundary splits, numbering, character budgets — but the best threads are split with intent, and it's worth editing the output rather than shipping it raw. Two principles cover most of it. First, one idea per post. If a part contains a claim and its evidence and a caveat, consider whether the caveat deserves its own post or deserves deletion. Second, end each post on a beat. A post that ends mid-argument leaks readers; a post that ends on a complete, slightly unresolved thought pulls them to the next one. Cliffhangers are the cheap version of this; the expensive version is making each post genuinely satisfying while still implying there's more.

Watch the joints, too. When the splitter breaks between two sentences that were doing one job — a setup and its punchline, a question and its answer — merge them back and split somewhere else. Automated splits are correct; they're just not always right.

The first post is 80% of the work

Only your first post appears in the timeline. Everything else is behind a click, which means the opener is not an introduction — it's an advertisement for the rest of the thread. The classic mistake is starting with context ("I've been thinking a lot about hiring lately…") instead of payoff ("I've interviewed 200 engineers. Here are the 5 answers that instantly tell me someone is senior."). State the concrete promise, imply the size of the payload, and let the numbering do the rest. If you can't write a compelling opener, that's usually a signal the thread doesn't have a spine yet — the problem is the material, not the marketing.

Numbering, length, and endings

Number your posts. "3/7" tells a reader mid-thread where they are and how much is left, which measurably improves completion — people finish things they can see the end of. On length: 3–7 posts fits most single ideas; 8–15 works for genuine step-by-step material; beyond that you're writing a blog post and should link one instead. And end deliberately. The last post is your highest-intent audience — they read the whole thing — so give them one thing to do: follow you, reply with their experience, or read the long-form version. One ask, not three.

Posting it: reply chains and pacing

Post the first part, then reply to it with the second, and so on — replying to your own post is what creates the thread. Do it promptly; a thread assembled over an hour can get orphaned engagement on the opener before the payload exists. If a part flops a link, remember every URL costs 23 characters of the budget. And resist the urge to delete-and-repost a mid-thread typo hours later: breaking the reply chain kills the thread's navigation. Fix typos in the first few minutes or let them live.