X spent three years building a machine that paid people to be popular, then spent one week admitting the machine was broken beyond repair. On September 7 the Creator Revenue Sharing program that made stars out of reply-guys and bot networks alike shuts down for good. A new program called Original Content Rewards opens the next morning, built to pay for the opposite of what the old one rewarded.
What did X actually change on September 7?
It stopped paying for engagement anyone could fake and started paying for impressions only a paying subscriber can generate. Revenue Sharing launched in July 2023 on a simple premise: ad revenue attached to replies on your posts, split with you, once you cleared a rising bar of impressions, fifteen million a quarter at first, then five million within weeks, then whatever kept the math working. By November 2024 X had already quietly swapped the funding source, moving payouts off ad revenue entirely and onto a cut of Premium subscription income tied to how verified subscribers engaged with your posts. The program that dies on September 7 was not the program that launched in 2023; it had been rebuilt once already, and it still did not survive. Original Content Rewards, opening September 8, narrows the target further: qualified impressions from Premium users on the Home Timeline, original posts only, reposts and aggregation excluded, an eighteen-plus age floor and a five hundred follower minimum.
Why does a platform kill the metric it invented?
Because the moment a metric decides who gets paid, someone builds a machine to hit it, and the metric stops meaning anything. Reply-ad revenue rewarded reply volume, so automated networks scripted replies, likes and reposts at industrial scale to farm the threshold. Musk said the quiet part out loud when he complained the system rewarded "spammy" accounts while genuine creators earned pennies, a strange thing to say about a system he built and ran for two years before noticing. This is Goodhart's Law running at platform scale: a measure that becomes a target stops being a good measure, and it takes a company exactly as long to learn this as it takes for the money at stake to become worth gaming. Watch the 48 hours before the September 7 cutoff for the tell: a spike in reposts and reply farming from accounts trying to bank one last payout on a system they know is about to disappear, the textbook shape of behavior facing an incentive about to go extinct.
Who is actually funding the creator you sponsor?
Not your ad budget, not since November 2024. The creator whose content your brand's ads sit next to on X is being paid out of Premium subscription revenue, a pool your media spend never touches. That decoupling matters more than the September reset itself: a brand director allocating budget to X on the assumption that ad dollars flow through to the creator layer is working from a model of the platform that has been wrong for nearly two years. Your reach and the creator's income now run on separate rails that happen to share a timeline.
A platform that changes what it pays for every twelve to eighteen months is not a media plan, it is a live experiment you are funding without a seat on the research team.
What should a brand director do with this before the next reset?
Treat every platform payout mechanic as a temporary policy, not a foundation, and build creator relationships that survive the next one. In practice that means:
- Put creator terms in direct contracts with a flat or performance fee you control, not a platform payout you are hoping trickles down
- Ask any creator partner what they are currently optimized to produce under the live payout formula, because it changed twice in three years and will change again
- Treat a sudden spike in a creator's platform-native metrics in the days before a known policy change as noise, not signal, when deciding who to renew
- Diversify creator budget across platforms with different reward architectures so one company's incentive redesign cannot reset your entire influencer program overnight
X did not fail its creators by paying them too little. It failed them by building a reward system smart enough to be gamed and dumb enough to take two years to notice. The lesson travels well past one platform: any KPI you attach money to will be reverse-engineered by whoever needs the money, and the only defense is knowing that going in.
Want this kind of thinking on your brand?
We build brand strategy, AI content and performance for the AI era.
START A PROJECT →