9x the conversion rate, 1/100th the revenue on note

Put the paid-conversion rates of two note writers side by side and the numbers point the wrong way. One has 1,500 followers and 8 paying subscribers, which is 0.5%. The other has 831,000 followers across social platforms and 478 paying members, which works out to 0.06%. Judged only by what share of an audience converts, the smaller operator is nine times more efficient.

The money runs the other direction. The subscription revenue is roughly 8,000 yen a month after fees for the first and an average of 772,500 yen a month for the second. The sources do not treat platform fees the same way, so the comparison only holds at the level of orders of magnitude, and there the gap is two. The side winning the rate by nine times is the side earning one hundredth of the money.

Of the 480 published cases, six report real figures for what they charge on note. Three of them go into one frame here: what was sold, where the buyers came from, how the charging was structured. The selection rule and the three cases left out are disclosed at the end.

Same shelf, different ways of charging

All three sell writing, and all three sell it on note. The only things that differ are the size of the audience and the form of the charge. Take them one at a time.

0.06% of 831,000 people becomes 9.27M yen a year

The operator publishing under the name AI FREAK reports 9.27M yen of revenue for the 2025 calendar year. The starting point was 38,500 yen in January of that same year and 65,680 yen in February. A month above one million yen arrived later, the best month reached 1.6M yen, and October through December was flat.

The shape of that money is not one-off paid posts. The creator page shows a 980-yen plan, a 1,980-yen plan and 478 members. If all 478 paid 980 yen the monthly total would be 468,440 yen, and if all of them paid 1,980 yen it would be 946,440 yen. The 772,500-yen monthly average implied by 9.27M yen over twelve months lands inside that range. Most of the annual figure is explained by the balance of live contracts rather than by which posts happened to sell.

What the operator writes about free posts is the other face of that structure. Material good enough to charge for went out on the free side without being held back. For one-off sales the faster move is to lock a post and collect immediately. Subscriptions do not work that way. Nothing starts until a reader decides the paid side is also worth reading, so free posts act as samples that cost nothing to distribute. The refusal to use hype in the copy comes from the same place. This is a business where revenue depends on people staying satisfied after they buy, not on closing a sale once.

The flat stretch from October to December is the same property showing through. Once new contracts and cancellations balance, the balance stops moving no matter how much gets published.

0.5% of 1,500 people loses 80% when three people leave

The second case writes about AI use and productivity under the name Kakeru. Before opening the magazine, note revenue was around 80,000 yen a month from an online course and single-article sales, against 1,500 note followers. On top of that came a subscription magazine at 500 yen a month, updated twice a week.

In the first month, 8 people subscribed within 24 hours of opening and almost nobody joined after that. The number taken on day one became the number for the month. The followers who act the moment they see an announcement are used up immediately, so subscribers stop growing unless new followers arrive. The promise of two updates a week, meanwhile, comes back every week as an obligation to write eight posts a month.

In the second month three people cancelled and five remained. Revenue fell from about 16,000 yen to about 2,500 yen. Three departures erased more than 80%. The reasons the writer gives: the promised twice-weekly schedule slipped to once a week and then ran late, the novelty of the launch wore off, and readers began noticing that the paid material was close to what was already free. That last one is where subscriptions are exposed. A one-time sale closes when the money changes hands, while a subscription puts that difference back on trial every month.

The third month was a repair job pointed entirely at subtraction. The theme narrowed from information about AI to methods the writer had tested personally for earning more than 50,000 yen a month, updates were fixed at once a week on Friday, the five remaining subscribers got an apology and a preview of the new plan, and the price went from 500 yen to 700 yen. Subscribers came back to 15 and revenue reached about 8,000 yen.

The numbers deserve to be read plainly. Fifteen subscribers at about 8,000 yen does not beat the first month of eight subscribers at about 16,000 yen in money. What recovered was the headcount, not the revenue. The writer puts each post at two to three hours and the month at eight to twelve hours for four posts, which divides into about 8,000 yen as an hourly rate between 600 and 1,000 yen, and calls it a poor trade.

The way the price increase worked is the interesting part. The content narrowed, the frequency dropped, the price went up, and the headcount still grew. The people paying had been replaced. A reader waiting for general AI news and a reader who wants a tested method for 50,000 yen a month are different buyers, even when both sit inside the same 1,500 followers.

One-time sales do not ride on audience size. They ride on volume

The third case, Majima, sells one-time purchases on note while holding a day job. Four earlier attempts were abandoned, and those figures sit in the same post. Mercari resale: 30,000 yen of first-month revenue against 8,000 yen of profit, dropped after two months. Blogging: two hours a day for six months, 100 pageviews a month, zero revenue. YouTube: more than five hours per video across ten videos, 20 subscribers, dropped after three months. Crowdsourced writing: 3,000 characters in three hours for 1,500 yen, an hourly rate of 500 yen, dropped after one month.

On note, 500 yen of revenue arrived on the third day after the first post, and more than 400,000 yen a month after five months. The sentence that matters more than the total is that a post written five months earlier still produces 20,000 to 30,000 yen a month. If one post keeps producing that, a dozen or so at the same level reach the 400,000-yen range. This revenue is a function of how many posts exist, not of how large the audience is.

There is no cancellation in a one-time sale. The transaction closes at purchase, so no monthly renewal decision ever arrives. In exchange, revenue that has already landed only grows by being stacked on. That is a different shape of growth from a subscription where 478 members are still there next month.

This is also the case with the thinnest evidence of the three. No dashboard showing 400,000 yen a month appears in the free portion, and both the number of posts and the price per post sit behind the paywall. The editorial desk treats this 400,000 yen as unverifiable self-reporting. What holds up structurally is the speed of the first 500 yen on day three and the persistence of 20,000 to 30,000 yen a month from a five-month-old post.

What set the revenue was never the writing. It was the variable the charge rides on

Subscription revenue is audience size times conversion rate times price. 0.06% of 831,000 people is 478 members, and 980 to 1,980 yen against that produces a band between 460,000 and 940,000 yen a month. 0.5% of 1,500 people is 8 subscribers, and nothing you multiply that by lands above the tens of thousands of yen. A conversion rate nine times better cannot repair a denominator 554 times smaller, and it did not.

A small audience also leaves a single-digit numerator. Three cancellations taking 80% of revenue measures no one’s effort. It is a property of the number 8. No campaign fixes that fragility.

One-time revenue is volume times price. The audience term is absent, so the question of how many followers someone has does not even apply. Five months to 400,000 yen a month is explained by how many posts were written and what they cost, not by how many readers were watching.

Having put the three side by side, the editorial desk reads it this way. What set revenue on note was not the quality of the prose but how many buyers already existed outside the platform and which charging format was pointed at them. The long-standing reading here is that a price is set by who the buyer is rather than by what the seller built, and this is the headcount version of it. The same writing moves a different order of magnitude when the buyer count moves three.

Speculating from here: whether subscriptions suit someone has almost nothing to do with their writing. It has to do with how many people they gathered elsewhere before charging anything. Of these three, the only one who raised an order of magnitude through subscriptions is the one who held YouTube, Instagram and TikTok outside note.

All three had a three-digit hourly rate at some point

One more thing the three share is the hourly rate. Kakeru is at 600 to 1,000 yen an hour for four posts and about 8,000 yen. AI FREAK’s 9.27M yen looks like 10,000 yen an hour when divided by the roughly 900 hours of two to three hours a day for 365 days, but that is a figure available only after running the full year. At January’s 38,500 yen the same work paid about 400 yen an hour. The crowdsourced writing Majima did before note paid 500 yen an hour.

Line up only the figures from after the jump and that stretch disappears cleanly. In all three cases, the hourly rate at the start was in the hundreds.

Why these three, and the three left out

The selection was made here, so the rule goes on the record. Of the 480 published cases, six disclose real figures for note earnings. From those six, the three taken are the ones where the charging format is unambiguous and where the origin of the buyers is described by the operator.

The three left out have reasons. One paid post, 80M yen over five years is a larger order of magnitude than any of these, but it is self-reported and includes downstream revenue, which leaves little room for verification. Kyoichi Anzai’s four years starts at 10,000 yen a month with one-time posts and passes one million yen a month in year three when a subscription magazine opens, which makes it a case about sequence rather than one to set beside others. That sequence, building readers through one-time sales before moving to subscriptions, points the same direction as the reading above. Ikeda Hayato’s 4.05M yen month reports note’s 1.07M yen as one line inside a blog revenue breakdown, so it is not a record of how the note side was priced.

Four limits apply. The 831,000 figure is the sum of YouTube, Instagram and TikTok followings, overlap unknown, and it is not a count of note followers, while the 1,500 is a number from inside note, so the nine-times comparison of conversion rates is not strict. The sources do not treat fees alike: the first month’s roughly 16,000 yen is published alongside about 11,200 yen after fees, and later months appear net of fees, which is why money here is compared only by order of magnitude. All three are people who wanted to publish their figures, so anyone who ran the same subscription without gaining members and quietly closed it is absent from those six. And with n=3 there is no median.

For distributions, the ceiling of subscription revenue measured through churn sits in ten cases that published a churn rate, and monthly revenue across solo and side businesses sits in how far one person can go. Definitions of the money terms are in the data page.

The name of the platform explains nothing about the shape of the revenue

After writing these three, grouping cases by platform name has lost credibility here. 9.27M yen a year on note, 400,000 yen a month on note, 8,000 yen a month on note. Lined up, they say nothing beyond sitting on the same shelf. What set the shape of the revenue was the charging format and the number of people waiting outside before any charge existed.

A headline reading “I made X on note” therefore carries the money and nothing else. Whether that money rides on audience size or on volume decides how next month grows and how badly three cancellations hurt, and those are different businesses. That is the part to look at before the amount.