- Growth Bytes
- Posts
- X deleted the program
X deleted the program
We got monetized four days before they announced it. Here is what replaces it and why the new bar is ten times lower.
Web3 Matters
August 10, 2026
AI + Web3 intelligence for builders and creators
Growth Bytes
Issue #012 · August 10, 2026 · theshipguild.com
This week: Trav hit five million impressions, got monetized, and X paid him sixty dollars. Days later X announced it is deleting the entire program. Here is what replaces it on September 8, why the new threshold is ten times lower, and why that makes it harder rather than easier. |
Section 1 · Livestream Recap
We Got Monetized Just in Time to Watch X Delete the Whole Program
From Friday's stream on using AI agents to grow on X, updated with this week's announcement
Last week we published the full reply strategy that took Trav from six thousand daily impressions to five million in four weeks. The account got monetized. The first check arrived. It was sixty dollars.
Then, days after we recorded that stream, X announced it is retiring Creator Revenue Sharing entirely. That sixty dollar check is now a relic of a program with a death date on it.
There is real opportunity buried in the announcement, and there is a catch that most of the coverage is glossing over. Both matter to you.
$60 What five million impressions actually paid | Sept 7 The day Creator Revenue Sharing is retired | 500K New threshold, down from five million |
The Transition Calendar |
Aug 7 | Enrollment in Creator Revenue Sharing closed. Already in? You keep going. |
Aug 14 | Final payouts begin. Three of them: Aug 14, Aug 28, Sept 11. |
Sept 7 | Creator Revenue Sharing is retired permanently. |
Sept 8 | Original Content Rewards opens. Everyone applies fresh. |
Sept 25 | First payout under the new system lands. |
What actually changes:
The entry bar drops by a factor of ten. Five million impressions becomes five hundred thousand over ninety days, alongside the same five hundred verified followers. If you looked at the old number and quietly gave up, this is genuinely your moment. The door just got a lot wider. |
But a qualified impression is a much narrower thing. It has to be a unique view, from a verified Premium subscriber, on the Home Timeline, with at least half your post visible on screen. Duplicates, paid, promoted and fraudulent impressions are all stripped out. Trav's five million raw impressions would convert to a small fraction of that under this definition. Five hundred thousand qualified is a harder number than it looks. |
Replies are excluded permanently, now in writing. What Trav found out the hard way after grinding to monetization is no longer a quirk of the payout formula. It is written into the program. Reply impressions do not count toward the threshold and do not count toward earnings. |
And the big one: your content now has to pass an originality standard. X named what qualifies. Original reporting, genuine analysis, self shot video and photography, original graphics and illustrations. The company said openly that this exists to end clickbait, recycled material, and content built purely to farm the payout system. For the first time, a human is not just moderating your account, something is judging whether you actually made the thing. |
⚠️ This Just Got Much More Serious AI Slop Is No Longer Just a Ban Risk. It Is a Disqualifier. On Friday we warned that X has suspended hundreds of thousands of accounts for AI generated slop and for agents replying on the user's behalf, and that some regions now let users report content specifically as AI slop. Days later X made originality a formal requirement to get paid at all. The warning aged well and it now has teeth. Recycled and machine generated content does not just risk your account, it fails the eligibility test. Using AI for research, angles and drafts is still fine. Letting it write and press send is now a direct path to earning nothing. A human making the actual thing is the product. |
One thing that did not change: replying is still the single strongest way to grow the audience that will eventually see your original posts. It just moved from being the thing that pays you to being the thing that builds the audience that pays you. The method is completely different from the volume approach that got Trav monetized, there is a 50/30/20 split at the centre of it, and the full document lands in the vault this week.
The read on all of this: X just made "post less, post better" the literal business model. Every part of the new program rewards a creator who publishes fewer, genuinely original posts that verified people stop and read. Trav spent four weeks grinding volume for a program that is now being deleted, and the replacement pays for exactly the opposite behaviour. If you are building something real rather than farming attention, this change is on your side.
"The negative consequences of breaking the rules are far worse than any upside from playing by them." Trav, on why the first step is always learning the algorithm |
One research tip worth the whole stream: when you ask an agent to study the X algorithm, it will confidently pull an outdated version. Trav's agent cited the January release and missed a more recent update entirely. It only admitted the gap when challenged directly. Always add this line to your prompt: "Confirm you are reading the most recent published version and tell me its date." Then push back on the answer.
Section 2 · Method Spotlight
The Fix for When Your Agent Suddenly Gets Stupid
Both of us had our worst AI week of the year. Here is what actually solved it.
Trav's agent spent the week grabbing the first obvious answer, presenting it confidently, and being wrong. Q hit his Codex cap, switched models, and watched his entire multi agent system fall apart with no explanation. If you felt this too, you are not imagining it.
Two things are probably happening at once. Providers have every incentive to make cheaper models feel worse so you upgrade, and separately, your own setup likely has weak points that only surface when a model changes underneath it. You cannot control the first. You can absolutely fix the second.
🪄 The Self Correction Protocol 1. Stop the task. Do not keep re prompting a failing agent. 2. Ask: "What is wrong with the way you are doing this? Give me a detailed response covering everything you are doing wrong. Think it through." 3. Then: "Keep this in your memory permanently and never make those mistakes again." |
That third line is the part people skip, and it is the only part that compounds. Without it you are fixing the same failure every week. With it, the agent gets permanently better at the thing it was bad at.
Set your enforcement rules once. Q's standing instructions to his agent, which are worth copying: |
Ask for anything you need from me (credentials, context, clarification) before starting the task, not halfway through.
Once started, do not stop until the task is genuinely complete.
Never finish without double checking your own work.
⚠️ The escape hatch matters. Trav told an agent "do not come back until you figure this out" on a simple animation task and it burned ninety minutes looping on something he could have done by hand in twenty. Persistence without a way to raise a blocker is not persistence, it is an expensive stall. Always pair "do not stop" with "ask me first if you get stuck."
On Grok: if you have a verified account you already have it, and it is the only practical way to get live X data without handing over credentials or paying for expensive API access. It is also sycophantic and loose with accuracy. Asked for his last twenty original posts, it returned his last twenty replies and built an entire strategy on the wrong data. Set explicit parameters, and always include the line: "If you do not know, say you do not know. Do not give me an answer designed to please me."
⚓ DROPPING THIS WEEK The Organic Growth Reply Document >The full method for post monetization growth, including the 50/30/20 rule, how to build your reply target lists, and the Grok audit prompts that actually return usable data instead of confident nonsense. > Hold any amount of SHIP, connect your wallet at theshipguild.com/rewards, and it unlocks alongside the full library. Read only verification works if you would rather not connect. |
🔥 ONE FAVOUR We are rebuilding this newsletter to be shorter and sharper, and we want it aimed at what you are actually working on. Hit reply and tell us one thing you are building or stuck on right now. We read every single one, and we will answer. If Growth Bytes earns its place in your inbox, share it with one builder who needs it. [INSERT BEEHIIV REFERRAL URL] 🐦 X: HERE Every Monday. See you Friday at 4pm UTC. |
Reply