The Distribution Loop Behind Postiz's $17K-to-$143K Month

Most coverage of Postiz credits the agentic pivot. The story goes: Neville rebuilt his social media scheduling tool for AI agents, and the market rewarded him — $17K to $143K in monthly recurring revenue in four months, a hockey stick in a category where most SaaS grows in single digits month over month. The pivot mattered. But a pivot to agent integrations doesn't move that number by itself, and treating it as the cause is how you end up copying the wrong part. What actually moved the number was a distribution loop: one user's article, a format decision, and a paid repost network. The pivot made the product worth paying attention to. The loop manufactured the attention.

The numbers, in the order they happened

Before the loop, the curve was unremarkable. It took Neville a year and a half to get Postiz to $21K MRR — the normal, grinding version of a SaaS. Then, per the revenue he reported publicly: $41K in February, $76K in March, $80K in April, $109K in May, $143K by late June. Roughly $1K of new MRR per day at the peak. In the same window, churn fell from the mid-to-high 20s to 13.7%. Both changes trace to the same few weeks — which is useful, because it tells you what the inflection point actually was.

The trigger was distribution, not product

The inflection point was an article written by a user, not by Neville. Oliver Henry, who was running his TikTok fully automatically with an open-source agent framework and Postiz, published a piece on X about it. It got 7.2 million views. Postiz's trials jumped to roughly 700 concurrent per day and stayed there; a second article added another 1.5 million views. The product didn't change that week. The distribution did — and the demand arrived before the pivot was even finished.

What Neville did that most founders won't

He didn't sit on the signal. Four moves, in order:

1. He wrote his own article about the same stack. Two hundred followers, half a million views. Then a second article, another half million.
2. He read the format shift correctly. X was surfacing long-form articles over posts — more dwell time, more algorithmic push — so he stopped treating articles as occasional and made them the channel.
3. He built a paid repost loop. He watched the activity pages of creators getting traction, saw who was buying paid reposts, and copied them: $60–$350 per repost, roughly twenty reposts per article shipped.
4. He measured MRR per article and scaled what worked. Not vanity metrics — revenue movement per piece of content.

The insight worth stealing isn't "write an article." It's that during a rising-topic window, attention on X is purchasable in small, repeatable increments — and few people are doing it properly, because buying reposts is still slightly embarrassing to admit. That embarrassment is the arbitrage.

The part case studies skip: the pivot halved churn

The agentic change did its real work on retention, not acquisition. Churn went from the mid-to-high 20s to 13.7% — not from a pricing change or a retention campaign, but because the product changed who does the work. Before, a customer had to open Postiz daily, write, iterate, post; when that routine lapsed, they cancelled. After, an agent does the daily work on schedule, without intervention. The thing that cancels a subscription is a human who stops doing a human thing. An agent doesn't stop. Same thesis as the growth side, applied to the number that compounds.

What's actually copyable

• Build the surface before the trend, not during it. Postiz had the API, the integrations, and the agent-ready docs before "agentic" was a marketable word — and it was listed in Claude and ChatGPT's marketplaces when the wave hit. The people who build when the trend peaks are the ones the trend eats.
• Budget for distribution, not just production. $60–$350 per repost is a rounding error next to agency retainers. Most founders spend the money on making, not on getting seen.
• Reliability is a growth feature. Neville stopped shipping new features on purpose — thirty channels, all of them have to work. His reasoning: the apps agents choose are the ones that don't break, and agents can tell which those are.

The failure mode, and the gap

The uncomfortable part of this story is how many people watched it happen and ran nothing. The hosts interviewing Neville said it on air: they saw the viral articles, thought "I should do that," and didn't. The information was public, free, and in their feed. What they lacked wasn't insight — it was a loop: a standing system that watches a surface, writes for it, distributes it, measures it, and repeats. Neville ran that loop by hand for four months, and it consumed his calendar. That's the specific gap Collision is built to close: research, writing, distribution, and learning running as one system with one memory, so the Postiz loop runs without the founder becoming a full-time distribution machine.

Source: Starter Story interview with Neville (youtube.com/watch?v=n7Obxh9-Rc8). Revenue and churn figures as reported in the interview; not independently verified.