Twelve From OneTurn the post that worked into twelve more

threads skill

You post something. It goes further than usual. You feel good for an afternoon, and the next morning you open a blank box and start from nothing. That is the part that makes no sense. The post that worked is the only piece of real information you got all month, and almost everyone throws it away by lunchtime.

What this is. A skill you install in Claude. You paste in one post that performed and answer four questions. It hands back twelve posts, ready to publish — each one in a named shape, aimed at a named job, with a day to post it on. One in, twelve out.

the whole idea

A post that worked holds two assets. Most people keep one

When something outperforms, the instinct is that the words were good. Sometimes they were. But every post that lands is carrying two separate things, and they multiply independently.

The insightWhat it said. The thing people recognised and agreed with hard enough to react.

The skeletonHow it said it. The opening move, where the line breaks fall, where the turn comes.

Both work on their own, and the data shows each doing it. One creator took a single observation — avoiding a message because reading it makes it your responsibility — and ran it through two different openings: as an all-caps question it did 2,697 times her median, and as a “wait, so apparently…” discovery eighteen days later it did 1,028 times. Same insight, different skeleton.

Another built everything on one skeleton: “During a job interview, if they ask [X]…” and ran it across ten subjects. Every single version landed between 300 and 1,656 times his median. Same skeleton, ten different insights.

You tested one arrangement of two assets that both travel on their own. There were eleven more arrangements you never wrote.

before it writes anything

It reads your post back to you first

The first thing it returns is not a post. It is four lines: what your insight actually was, what your skeleton actually was, which of the eight shapes you used, and what that shape measures at.

Then it tells you which of the two carried the post, and this is the most useful sentence in the whole session. If your shape sits above baseline, your skeleton is an asset and worth reusing. If it sits below — and every list-shaped post does — it says so plainly: this worked in spite of its structure, not because of it.

Good news when that happens: your idea was strong enough to carry a weak frame. The frame is the one thing not worth repeating.

And if the post matches no shape it knows, it says so rather than inventing a label that sends the whole session the wrong way.

the measured part

Eight shapes, two jobs

Everything in the skill is measured against 22,214 Threads posts from 33 creators, August 2024 to September 2026. Reach is the share of posts in that shape that beat their own author's median by eight times or more, on a baseline of 4.96%. Replies is median replies-per-like against a baseline of 0.056.

Shape	Reach	Replies	Posts	Best at
Normalize	2.9×	0.77×	171	reach
Callout	2.8×	1.39×	643	both
Comparison	2.4×	0.92×	109	reach
Flat truth	2.3×	0.87×	2,345	reach
Stop / Don't	2.1×	0.76×	188	reach
Question	2.0×	4.59×	368	replies
If you…	1.9×	1.02×	321	balanced
The Hill	1.3×	—	121	reach, mildly

Two things worth taking from that table even if you never install anything. The Question is a different instrument — four and a half times the normal reply rate while being middling at reach, so using it to chase views wastes it. And the Callout is the only shape that does both jobs well.

And the shapes that lose. The four list shapes — HEADER: then short lines, a list of three, a list with a closing reveal, a negative list — all land below the 4.96% baseline, at 0.4 to 0.7 times. They are the shapes people reach for most and the weakest in the set. A post that went big as a list went big on its idea, and the list was a tax it paid.

how the twelve are chosen

A grid, not a list of tricks

This is the part that decides whether you get twelve posts or one post wearing twelve labels. It builds a grid: rows are what you talk about, columns are the shape you say it in. Twelve posts means twelve different cells.

6 · your original insightThrough six shapes it has not used yet. Same claim, six different openings.

2 + 2 · two new subjectsOther things your people are stuck on, each through two shapes. These come from the fourth intake question.

1 · the mirrored groupSame insight, the adjacent audience. Women and men, beginners and veterans, the ones who left and the ones who stayed.

1 · the original, unchangedCopy and paste, scheduled last. The one people argue with, and the one with the best evidence behind it.

Two caps keep it honest. At most two of the twelve may share your original's skeleton, so you never get six posts that all read [X] is cheaper than [Y]. And no shape appears more than twice, with the second one built differently from the first. Then it reads all twelve as a flat shuffled list, out of grid order, because duplicates hide across rows — and replaces any two that could be mistaken for each other.

Eleven distinct posts beat twelve with a repeat. The skill is told that explicitly, which is why it will sometimes hand you eleven.

the counterintuitive one

Posting it again, unchanged, beats improving it

Among similar post pairs by the same author, both versions hit 23.8% of the time when the text was identical — against 19.9% when it had been rewritten. Republishing unchanged beat republishing improved.

The clearest case in the set: the same post published on 1 July at 1,803 times its author's median, again on 14 July at 1,044 times, and again on 17 August at 4,817 times. The third run was the biggest of the three.

Across the whole set, twenty of the twenty-five measured accounts republish identical text. It happened 1,131 times, and in 304 of those cases two or more copies each went viral independently.

Almost nobody saw the first one. A post reaches a slice of your followers and a smaller slice of everyone else, so the second run is a first look for most people.

There is a limit and nobody in this data found it. One creator ran the same sentence four times: 244×, 208×, 71×, then 2×.

when to post them

Closer together than you think

Among similar posts by the same author, how often both versions hit, by the gap between them:

Gap	Both hit
0–1 days	32.0%
2–7 days	29.5%
8–21 days	25.0%
22–60 days	14.2%
60+ days	8.1%

Which is the opposite of the usual advice about spacing things out. The twelve come back spread across ten to fourteen days, each with its day already on it. The mirror goes out the same day as the post it shares an insight with; the unchanged re-run goes last.

One honest caveat, which the skill will tell you too: creators on a hot streak publish variants close together and everything of theirs hits, so part of that 32% is the streak rather than the spacing. The direction holds. The effect is softer than the number looks.

what it asks you

Four questions, no homework

Paste the post that worked. That is the whole first message.
Roughly how did it do next to your usual? A bit better, a lot better, or best thing you have posted. Three options, nothing to go and look up — and each answer changes the mix you get back.
One line: who do you help, and with what? It is not a positioning exercise. One line.
Name two or three other things those people get stuck on. This is the one that does the real work. These become the subjects for a third of the output.

Skip the last two and it carries on anyway: it takes the subjects from your post's own territory and marks those slots guessed subject — swap in your own. A slot you fix in ten seconds beats a question that stalls the session.

take it

Download

post-multiplier.skill Claude skill · 13 KB Download

This one needs Claude. It is a skill, which is a thing you add to Claude so it knows how to run a specific job properly. It will not work in ChatGPT — the file is instructions written for Claude to follow. The free plan is enough.

five minutes, once

How to install it

All in the browser. Nothing to install on your computer, no terminal, no code.

Download the file above. It lands in your Downloads as post-multiplier.skill. Do not try to open it — it is not meant to be read, it is meant to be uploaded.
Go to claude.ai and sign in. A free account works.
Open Settings, then Capabilities, then Skills. The wording moves around a little as they update things — you are looking for the Skills section in your settings.
Upload the file. It appears in your list of skills and stays there. You do this once, ever.
Open a new chat and paste your post. “This one did well, what else can I do with it?” is enough.

If it doesn't start on its own, name it: “use the post multiplier”. Works every time.

what comes back

Twelve posts, written, with a job and a day on each

Not twelve briefs telling you to write something. Twelve posts you could paste as they are. Each one arrives like this and with no commentary underneath, because the shape, the job and the day are the commentary:

4 · Question · replies · day 5

[the post, in full, ready to publish]

Ten of the twelve come in under fifteen words, because in this set posts of fifteen words or fewer hit 9.98% against 1.92% for posts of 41 to 70. The two exceptions are the ones that keep your original's skeleton, which keep its length too. All of them are text only: text-only posts in the set average 242 likes against 170 for posts with media.

And where your post gave it something worth keeping, it lifts your phrasing rather than paraphrasing you. Your words already proved they travel.

so you are not disappointed

What it does not do

It will not turn your post into a carousel, a newsletter, a video script or a thread. None of those are in the measured set, so anything it said about them would be invention, and it is built to say so instead.

It will not write you a sales post either. Nothing in the data can see a sale, and a slot with no number behind it does not belong next to eleven that have one.

It also will not tell you whether your post deserved this. One of the twelve may well beat it, which is rather the point.

The 7-Day Views Boost More free resources