Three months of real numbers from a paid note magazine: subscribers 8 → 5 → 15, revenue ¥16,000 → ¥2,500 → ¥8,000

Subscription case studies usually only show you the numbers going up. What makes this one unusual is that the full three-month arc (growing, shrinking, and growing again) is disclosed with both subscriber counts and post-fee revenue for every month. And the amounts involved are small: about ¥16,000 in the best month, about ¥2,500 in the worst. It’s a direct look at how much a recurring-subscription magazine actually returns for someone with roughly 1,500 followers.

The writer goes by “Kakeru” and posts about AI tools and productivity. Before launching the magazine, his situation was: 1,500 note followers, and roughly ¥80,000/month in note sales (from a course and one-off sales). This is the record of the three months he spent trying to add a recurring-revenue pillar on top of that.

The three-month trajectory

He also gives the subscription rate: 8 subscribers in the first month against 1,500 followers is 0.5%. When estimating how a subscription magazine will take off, that 0.5% is one useful real-world data point.

Month one: hitting a ceiling within 24 hours

8 people subscribed within the first 24 hours of launch. So far, so smooth, but the problem is that growth was “essentially zero after that.” If the segment of existing followers who act immediately on an announcement all convert on day one, subscriber growth stalls unless new followers keep flowing in. Day-one numbers essentially became the month’s numbers.

Meanwhile, the cost didn’t stop after day one. A commitment to update twice a week arrives every week as an obligation to write eight pieces a month. Against ¥16,000 in revenue, that workload starts to feel disproportionate, a feeling that fed directly into the collapse in month two.

The dark turn: what happened in month two

In month two, subscribers dropped from 8 to 5, 3 cancellations. Revenue fell from about ¥16,000 to about ¥2,500, an over 80% drop in real terms. The writer breaks the cause down into three things.

One was the breakdown in update frequency. What had been announced as “twice a week” became once a week, and then fell further behind schedule. The second was that the “newness” of launch had worn off. The third, and most fundamental, was that readers started noticing “this isn’t much different from what’s free on Twitter.”

This is the crux of the recurring-subscription model. With a one-off sale, the transaction is complete the moment someone buys. A subscription requires the reader to make a renewal judgment every month, so the moment it becomes apparent that the free posts and the paid content are the same thing, that gap gets clawed back in the form of cancellations. Run with the boundary between free and paid left blurry, and it gets tested again every time the month rolls over.

The turnaround: cut, narrow, raise the price

Four moves were made in month three, and all of them pointed toward “narrowing,” not “growing.”

First, narrowing the concept. He tightened the broad theme of “AI tool information” down to “methods I’ve personally tried and confirmed work, for earning ¥50,000+ a month.” That’s a shift from generic information to firsthand, verified information.

Second, cutting the update frequency from twice a week to once (Fridays). Rather than keep promising a cadence he couldn’t keep, he lowered it to one he could actually meet.

Third, he apologized to existing subscribers and sent them a preview of the new format, explaining himself first to the 5 who remained.

Fourth, he raised the price from ¥500/month to ¥700. Raising the price right after subscribers dropped runs against instinct, but since he had changed the nature of the content, this reflected that change in the price too.

The result: subscribers went from 5 to 15. 13 new subscribers came in, and post-fee revenue came to about ¥8,000.

Breaking down what actually worked

The most reasonable read is that the main driver of the recovery was the shift from “content also readable for free” to “content not readable for free.” Since the cause of the cancellations was “no different from Twitter,” the fix had to be differentiation, not frequency or price. The month-three concept change addressed that cause head-on.

Cutting the update frequency also worked, paradoxically, in revenue’s favor. Content that’s thin twice a week is worse for retention than content that’s dense once a week. The “psychological burden of deadlines” he mentions himself is roughly halved when it goes from eight pieces a month to four.

In this context, the price increase functioned not as a standalone tactic but as a declaration of the concept change. Keeping the price at ¥500 while only changing the content internally would have made the change harder for existing followers to notice.

That said, looking at the numbers coolly: 15 subscribers at about ¥8,000 doesn’t exceed the ¥16,000 (pre-tax, pre-fee) from month one. What recovered was the subscriber count, revenue never got back above its opening month. Calling this arc a “V-shaped recovery” isn’t accurate.

Does it actually pay off?

The writer runs his own hourly-rate math. Each post takes roughly 2–3 hours to write; at four posts a month, that’s 8–12 hours. Dividing roughly ¥8,000 in revenue by that comes to an hourly rate of ¥600–1,000. “Honestly, it’s not a great deal,” he writes.

Even so, he lists three reasons for continuing: there’s still room to grow, he gets feedback directly from subscribers, and the weekly deadline improves the quality of his output. This three-month recurring magazine has been repositioned not as a revenue business in its own right, but as an investment in a customer touchpoint that supports his existing note sales (roughly ¥80,000/month).

He also lays out who it fits and who it doesn’t. It suits people who can treat a deadline as a challenge, who value community-style connection, and who work in niche topics. It doesn’t suit people for whom deadlines are stressful, people expecting rapid growth, and people already selling high-ticket products. That last point matters: if you’re already selling a course worth tens of thousands of yen, the rationale for spending time on a ¥700 magazine is thin.

How far does this generalize?

What’s replicable is the order of operations when things slow down. When cancellations start, the instinct is to reach immediately for price and frequency, but this case addresses the actual cause (“difference from free content”) first, and only adjusts frequency and price afterward. Because the relationship between cause and countermeasure is explicit, the same sequence should hold up in other topics too.

What’s not replicable, or is conditional, is scale. With a base of 1,500 followers and a 0.5% subscription rate, the absolute subscriber count is small enough that 3 cancellations can wipe out 80% of revenue. This fragility can’t be solved through tactics, only by growing the base. Conversely, for a creator with a sufficiently large base, the “dark turn” that happened here would likely show up as a much gentler wobble in the numbers.

And one more thing: the effect of publishing this record itself is unmeasurable. If the timing of the 13 new subscribers in month three overlaps with the timing of publishing this operational writeup, the increase might be attributable to the act of publishing itself rather than the concept change. The disclosed data doesn’t allow that distinction to be made.

Related reading

• 24 Kindle books — a different pattern of an individual stacking up content sales
• Plausible Analytics — a contrasting case of scaling a monthly-billing model up to roughly ¥100M in annual revenue