A film nobody sees did not launch
The film is one asset. The launch is what reaches people on the day. This is how we run the second half of the job, and why it is planned before the edit is locked.
Distribution is half the work
A launch is not a publish date, it is a competition for one hour of attention. We plan the posts, the roster and the timing before the edit is locked, because that hour is where the outcome is decided.
How the X algorithm works
Four things decide how far a post travels. A launch plan that ignores them is a guess.
01 of 04
01
Your followers see it first
The post goes to a slice of the people who already follow you. Nothing else happens until that slice reacts, which is why a good film on a quiet account stays quiet.
Nothing reaches the right side until the left side reacts. 02
Half of every feed is out of network
X puts the For You feed at roughly half in-network, half out-of-network. That second half is where a launch reaches people who have never heard of the company, and it is only unlocked by scoring well with the first half.
Roughly the split X describes for the For You feed. 03
Replies count for far more than likes
The ranker weights a reply at twenty seven likes, and a reply you answer at a hundred and fifty. The brief is not to collect likes, it is to say something worth answering.
- Reply answered+75
- Reply+13.5
- Repost+1
- Like+0.5
04
The first hour sets the ceiling
Reach compounds off early engagement. Two identical posts, one that earns replies in its first minutes and one that does not, finish the day on different numbers, and the slow one never catches up.
Same film, same account, two different launches.
What the ranker pays for each action
X's own weights, from the recommendation code it open sourced in March 2023. This is why we brief creators to write a real take, and why we never buy engagement: the cheap kind is weighted near zero and the fake kind gets reported.
Earns reach
- Reply the author answers+75
- Reply+13.5
- Profile click that leads on+12
- Repost+1
- Like+0.5
- Video watched+0.005
Costs reach
- Not interested in this-74
- Report-369
Five steps to a viral launch
Half the launches in our archive land between 300K and 1M views, and the biggest have gone past 2M. This is the sequence behind them.
01
Anchor the launch on one post
Everything points at a single post, normally the founder's, with the film in it. It holds the claim and the link, and it is where replies, bookmarks and profile clicks accumulate. Spread those across five posts and you get five small launches.
Every quote points home. One post accumulates everything. The same spend, seven separate posts, nothing accumulates. 02
Pick the roster for the room
Follower counts are the least useful number in a brief. We pick accounts whose replies your buyers read, and you see every name and price before anything runs.
We buy the overlap, not the biggest circle. 03
Quote tweets, not reposts
A repost dies in a timeline as your post. A quote is a new post with its own audience, its own replies and its own score, and it still sends people back to the anchor. Every quote is written for the account posting it.
One audience, reached twice, in your words. Two audiences joined, and a link home. 04
Front-load the first hour
The opening quotes land inside the first hour, close together, because that is where the ceiling is set. The rest are spaced across the day and into the next: eight posts in ten minutes reads as an ad, eight across two days reads as news.
05
Hold something back for the week after
Launch day attention decays inside a day. A follow-up post, a cutdown or a result the founder can report buys a second rise in the same week.
A second rise, rather than one long fade.
Related examples
Launch posts that ran this way, with the film in the post.
Odyssey@odysseymlToday we’re unveiling Odyssey-3, a big step forward for foundation world models. It can control robots, power humanoids, drive cars (on the roads of India!), train AIs, pilot drones, and even play video games. We can’t wait to see what intelligent systems it enables.
1:36Sep 15, 2026
- 2.2M+
- Views
- 4,412
- Likes
- 343
- Reposts
- 258
- Replies
Jims@JimsYoung_Most "agents" are basically interns with API access. They ask you to subscribe to the Times. Download the file. Forward it over. You're not deploying an agent at all. You're managing one. It's why we built Anyway. Anyway is the financial OS for agents that actually operates by
1:41May 28, 2026
- 2M+
- Views
- 1,334
- Likes
- 5,033
- Reposts
- 125
- Replies
Ridge@joinridgeIntroducing Ridge Forge! For the past year, our autonomous agents have been quietly at work for SaaS founders and AI startups: tracking funding rounds, product launches, and hiring surges, resolving the right contacts, and booking meetings at scale.
2:08Aug 12, 2026
- 470K+
- Views
- 205
- Likes
- 18
- Reposts
- 30
- Replies
Dipanjan Dey@Dipanjan_DeyIntroducing Kombai 2.0 - the first AI design engineer. We keep hearing that AGI is almost here. Still, we’re stuck with coding agents that don’t have taste and design tools that don’t understand our codebase. Both are artifacts of a world where design and engineering were two
1:24Jun 2, 2026
- 440K+
- Views
- 409
- Likes
- 66
- Reposts
- 97
- Replies
Shensi Ding@shensiWe just beat the best frontier models at a quarter of the price. Now you can too, in one API call. Today we're launching Merge Fusion. Send one prompt to a panel of models, and a judge writes one answer better than any of them alone. Fusion beats Fable 5 outright at 1/4th the
0:45Jul 23, 2026
- 410K+
- Views
- 364
- Likes
- 56
- Reposts
- 76
- Replies
OsamaAtwi@osaAtwiToday, we're officially launching Kyrall Until now, AI models couldn't capture design intent in engineering. We're changing that. Upload specification documents, sizing tool results, requirements, and past designs, and Kyrall builds a Knowledge Graph, so it not only
1:47Jul 6, 2026
- 390K+
- Views
- 527
- Likes
- 34
- Reposts
- 70
- Replies
What you get from us
The plan, before it runs
The anchor copy, the roster with every handle and price, the running order and the window each post lands in. Approve, cut or swap any of it.
Public counts, not a dashboard
Every placement is a post you can open. We report views, likes and quotes off the posts themselves, with the date they were read, so nothing is a number only we can see.
The launches that underperformed
Some waves land flat. You get told which placements did nothing and what we would change, because the second launch is where that knowledge pays for itself.
Next launch
Planning a launch?
Tell us what is launching and when. Thirty minutes is usually enough to know whether we are the right team for it.