TL;DR
Use AI for thread architecture — hook variants, arc beats, a payoff line — then strip its tells (same-length tweets, “let me explain,” 🧵). Feed it a thesis, an audience, a reference thread you admire, and a banned-phrase list. Since X open-sourced its Grok-era ranking in January 2026, the signals that matter are concrete: a reply the author then replies to is weighted +75 versus +0.5 for a like, and external links cut reach 50-90% for non-Premium accounts (as of June 2026). Write threads that pull replies, not links.
The task
You have an idea you want to put on X. You know threads work when they have a strong hook, build tension, and pay off. You also know AI-written threads have a specific stink: same-length tweets, “let’s dive in,” “this is the future,” and every fourth post a 🧵. The job is to use AI for structure without inheriting its tells.
What X’s ranking actually rewards (June 2026)
In January 2026, xAI open-sourced the Grok-era ranking stack (Home Mixer, Thunder, Phoenix, Candidate Pipeline) on GitHub, so the weights are no longer guesswork. Write to them:
| Signal | Approx. weight | What it means for a thread |
|---|---|---|
| Reply that the author then engages | +75 | Ask a real question; reply to early replies fast |
| Reply to your post | +13.5 | The thread should be answerable, not just admirable |
| Profile open then like/reply | +12 | Hook hard enough that people check who you are |
| 2-min dwell in the conversation | +10 | A genuine arc keeps people reading down |
| Retweet | +1.0 | Nice, but worth ~75x less than a sticky reply |
| Like | +0.5 | The vanity metric; barely moves ranking |
Two more June-2026 realities from the source code and creator data:
- External links are throttled. Non-Premium accounts posting an off-platform link see near-zero median engagement, and reach drops roughly 50-90%. Put links in a reply, not in the thread body, or skip them.
- Premium reach is real. Verified (Premium) accounts average around 10x the impressions of free accounts. If you post threads seriously, the $8/month web tier pays for itself in distribution.
When AI helps — and when it does not
AI is good at thread architecture (hook variants, arc beats, payoff structure) and at compressing a dense idea into a 280-character constraint. It is bad at X voice, which is shorter than it thinks, less self-important, and meaner about its own openings. Give it a reference thread you love; ban its favourite tics by name in the prompt.
What to feed the AI
- The thesis in one sentence: what you actually want readers to take away
- Audience: who is in your replies if it lands
- A reference thread you admire (paste it, do not link)
- Banned phrases: “let me explain,” “the future is here,” “buckle up,” ”🧵,” emoji bullets, fake stat patterns
- Whether you want a CTA at the end (follow, sub, link, none)
- Personal anecdotes / data you can include. Without these the thread sounds like everyone else’s
Copy-ready prompt
Write a 10-tweet X thread.
Thesis (one sentence): [line]
Audience: [who]
CTA: [follow / sub / link / none]
Reference thread I admire: "[paste 5-10 tweets]"
Banned phrases / tics: [list, including "let me explain", "the future is here", "🧵", "buckle up", emoji bullets]
Personal anecdotes / data to weave in: [list]
Structure:
- Tweet 1: hook (under 200 chars, no emoji). Must be testable — read it alone and feel the pull to keep going.
- Tweets 2-3: tension (the obvious answer is wrong, here's why)
- Tweets 4-8: arc with one concrete example or data point per tweet
- Tweet 9: payoff (the line the reader will screenshot)
- Tweet 10: CTA — short, no salesy intent. Include one open question that invites a reply.
Constraints:
- Vary tweet length: at least 3 under 100 chars, at least 2 over 250
- No tweet starts with the same word as the previous one
- No "thread 🧵" indicator — let it speak for itself
- No external links inside the thread body
- Do not number tweets unless I asked
Output: the 10 tweets, plus 3 alternate hook variants for tweet 1. Mark which variant best fits the audience.For news / commentary threads, add: “Same structure, but tweet 1 must reference a specific real event in under 150 chars; tweet 9 must offer the contrarian read.”
Any current model handles this well. As of June 2026 the picks are GPT-5.5 (sharp, fast hooks), Claude Sonnet 4.6 (best at matching a pasted reference voice), and Gemini 3.1 Pro (long context if you paste a whole back-catalogue of your old threads as voice reference). For ready-made opener variants, see X thread prompts.
Thread vs. long post
Premium subscribers can publish a single post up to 25,000 characters instead of stitching 280-character tweets. Both can work; they reward different things:
- Thread keeps each beat answerable and bumps each time a new reply lands, which suits the reply-heavy ranking above. Better for discussion.
- Long post drives dwell time (the +10 two-minute signal) and is cleaner to read, but only the first ~280 characters show in the timeline — so the opening sentence still has to earn the tap. Better for a single sustained argument.
If you are unsure, default to the thread: it gives the algorithm more reply surface.
How to check the output is usable
- Tweet 1 stands alone: if you read only it, you would tap to expand
- Tweets vary in length (same-length threads are the AI-tell)
- Specific names, numbers, anecdotes appear, not “many people” or “a study showed”
- No banned phrase made it through
- The closing tweet asks something a real person would answer — that is your +75 signal
- Tweet 9 is the screenshot moment. Would you share it?
Common mistakes
- Weak hook tweet: most impressions stop at tweet 1
- Same-length tweets: visually a wall, scroll past
- A link in the thread body: throttles the whole thread’s reach (put it in a reply instead)
- Engagement-bait CTA (“RT if you agree”): short-term metric, long-term penalty
- Letting AI invent statistics: verify or remove every number
- Threads that summarise the blog post: see cross-platform repurpose; X is its own surface
FAQ
- How long should threads be? 6-12 tweets is the sweet spot. Longer threads need a stronger payoff to justify the scroll.
- Is ”🧵” still penalised? Not by a specific algorithm rule in the open-sourced 2026 weights, but culturally it reads as AI/marketer voice. Skip it.
- Should I post a thread or a Premium long post? Thread for discussion (more reply surface, which the +75 reply signal rewards); long post for one sustained argument and dwell time. Either way the first ~280 characters carry the click.
- Why are my link-in-thread posts dead? Since early 2026 non-Premium accounts posting off-platform links see roughly 50-90% less reach. Drop the link in a follow-up reply instead.
- Should I use the same prompt for Threads / Bluesky / Mastodon? Mostly yes, but each platform has different banned tics and no equivalent link penalty. Update the list per platform.
Related
- X thread prompts: focused thread prompt variants
- Cross-platform repurpose: turn the thread into other platform pieces
- Social caption AI: caption-only social posts
- Content calendar: schedule when threads ship
- Storytelling outline: story-led threads, not list-led
- Profile bio: bio that converts thread readers into follows