I've been running a daily analytics tracker on my X account for months. Every impression, every follow, every bookmark, every profile visit, logged and calculated down to the decimal. When I hit four consecutive payout cycles, I did something I hadn't seen anyone else do: I mapped every available metric against what I actually got paid.

The results surprised me. And if you're a creator on X trying to figure out what's worth your time, they might surprise you too.

Most creators assume the equation is simple: more impressions = more money. Post more, get seen more, get paid more. The reply guys operate on a version of this. Camp under big accounts, farm impressions off someone else's audience, collect a check.

I assumed the same thing. I was wrong.

Across my four payout cycles, I tracked a period where my impressions doubled compared to another period. Twice the eyeballs. The payout difference between those two cycles? Essentially zero.

Let me say that differently. I had one cycle where I was on fire. Viral-adjacent posts, huge bookmark numbers, new followers pouring in. And another cycle where things were quieter across the board. X paid me nearly the same amount for both.

When I ran the numbers across my three cleanest cycles, the payout variance was just 4.2%. Meanwhile, my impressions swung nearly 100% peak to trough. The formula, whatever it is, compresses output. More impressions didn't translate to more money in any meaningful way.

Two things tracked with payout direction across every cycle transition:

Replies received. Not replies I sent, but replies other people left on my posts. The only volume metric that moved with payout in both directions across consecutive cycles. Replies require actual time investment from your audience. Someone scrolling past and tapping a heart is low-effort. Someone stopping to type a response is high-effort. X appears to weight that difference.

Curiosity rate (profile visits as a percentage of impressions). This held in a remarkably tight band across all four cycles, and its small movements tracked with payout. When a slightly higher percentage of people who saw my content clicked through to my profile, payout moved with it.

I calculated revenue-per-unit for every metric I track. Most of them were wildly inconsistent cycle to cycle. But one stood out: revenue per post published.

Across my three clean cycles, the coefficient of variation on /postwas13.9/post was 13.9%. The next closest metric was $/reply at 19.6%. Everything else ( /postwas13.9/impression, $/follow, $/bookmark) was 28% or higher.

I'm not claiming X literally pays per post. But post count was the most reliable denominator in the equation. Showing up and publishing original content produced a more predictable revenue relationship than any engagement metric.

If your monetization strategy is built on farming impressions under other people's posts, this data raises a real question about whether that's actually paying you or just paying them.

The metrics that correlated with payout in my data were replies received (people engaging with your content), curiosity rate (people clicking through to your profile), and posts published (original content you created). All three of those require you to be the one posting, not the one commenting.

Reply-farmed impressions likely don't generate profile visits for you. They don't generate replies on your posts. And they don't count as posts published. If the formula weights any of those signals, and my data suggests it does, then every hour you spend in someone else's replies is an hour you're subsidizing their payout, not building yours.

And X seems to agree. As of this week, all aggregators had their payouts reduced to 60% this cycle, with another 20% deduction coming next cycle. The reasoning was explicit: flooding the timeline with stolen reposts and clickbait was crowding out real creators and hurting new author growth. Habitual bait posters are next on the list for permanent deductions.

The direction is clear. X is actively moving to reward original creation and penalize content farming. My data from four cycles already pointed that way. Now X is saying it out loud.

This is one account, four cycles, 56 days. That's enough to see patterns but not enough to make definitive claims. X doesn't publish its payout formula, so everything here is inference from observed data. Correlation isn't causation. My account operates in a specific niche (AI art and community) and these patterns might look different for a news account or a meme page.

And the algorithm is always changing. The aggregator deductions announced this week will reshape the payout landscape for the next cycle and beyond. What correlated with revenue in my first 56 days may shift as X adjusts its weighting. That's exactly why I'm going to keep tracking.

More cycles means more data points, tighter confidence intervals, and eventually enough information to start fitting a real model to the formula. I'll be documenting every shift as it happens.

But even with just four cycles and a platform actively rewriting its own rules, one signal holds steady: create, don't just react. The algorithm knows the difference. And now it's making sure the payouts reflect it.

Glenn is an Adobe Firefly Ambassador and AI creator documenting the craft of prompt engineering and creative process at @GlennHasABeard. He publishes The Render newsletter and creates the Stor-AI Time series adapting world folktales through AI-generated video.

Data methodology: Daily analytics tracking across 4 consecutive X payout cycles (56 days). Every available metric logged and calculated. Correlation analysis from observed data, not causal claims. One account, one niche. More cycles coming.