August 10-17: ¥43,280 in 45 days, ¥5,500 for five hours, and three honest zeros (2026)

Two of this week's better results came from tightening an existing loop. A Japanese creator says an AI-assisted website-audit service reached ¥43,280 in sales over the latest 30 days, within 45 days of starting that offer. Another spent the same five hours a month on Rakuten ROOM but moved half the time from posting to reviewing data; the reported monthly reward rose from ¥4,000 to ¥5,500. Neither report gives a startup-cost total, and the first figure comes from a paywalled preview. 1 2

The three zero-revenue cases are different in useful ways. GitPulse Weekly had 72 subscribers after 97 days but no paying customers. A Japanese note operator added followers while reporting ¥0 for week seven. Brief Work had a working marketplace, five registered accounts, and no revenue from strangers because its founder never personally asked a prospective client to post a brief. 3 4 5

The ranking below puts lower barriers first, then uses reported payout as the tie-breaker. "Cost not disclosed" means exactly that. It is not a $0 startup cost.

Quick scan

Lower barrier or smaller weekly commitment

Rakuten ROOM affiliate review: ¥5,500/month reported reward after a three-week change, with 5 hours/month of work; cash cost not disclosed. The fresh move was a 50/50 split between posting and data review. 2
GitPulse Weekly newsletter: $0 after 97 days, despite 72 subscribers and 14 issues; cash cost and weekly hours not disclosed. The first version was plain-text email through Gmail. 3
AI side hustle on note: ¥0 in week seven, with 28 followers and 20 total articles; cost, hours, and first step not disclosed. Follower count rose while views and likes fell. 4

Skill or distribution barrier

AI website-content audit: ¥43,280 in the latest 30 days, reached within 45 days; the author reports one hour per evening and does not disclose cash costs. The public preview says the work began with Claude-assisted product-page copy. 1
DevMarathon subscription software: $24 recurring revenue from one renewal; cost, hours, and first step not disclosed. The product has both developer and B2B hiring sides live. 6
Brief Work expert-gig marketplace: $0 from strangers after three months; build costs and weekly hours not disclosed. The founder spent three months building solo and did no outbound. 5

The six reports

1. Rakuten ROOM: the same five hours, moved to measurement

The operator has run Rakuten ROOM for a little over a year. The experiment was small and easy to describe: split a five-hour monthly schedule evenly between posting and reviewing performance data. The author says the change was observed over three weeks. The reported monthly reward moved from ¥4,000 to ¥5,500, while mediated sales rose from about ¥130,000 to about ¥180,000. The sales figure is merchandise flowing through referrals, not the author's payout. 2

The reported result is a 1.4x improvement in reward without more work time. That is a useful test for someone who already has a platform account and enough history to compare posts. It is not a clean new-account forecast: the starting asset was an account with more than a year of activity, and the source does not disclose fees, subscriptions, or profit.

First move: reserve 2.5 hours for posting and 2.5 hours for data review for three weeks. Keep the total at five hours a month so the result has a boundary.

2. GitPulse Weekly: 72 subscribers, no payer

GitPulse Weekly scans more than 500 trending GitHub repositories, turns that activity into commercialization ideas, and sends a report every Saturday. The creator started with plain-text emails through Gmail. After 97 days, 14 issues, and 72 subscribers, revenue was still $0 and there were no paying customers. The paid offer was a 13-page designed PDF, with Razorpay and Gumroad used for payment. 7

The distribution record is more useful than the subscriber count. Personal outreach and direct messages worked better than cold posts in Discord, Reddit, and Indie Hackers. Gmail also created spam-delivery problems, so the creator moved to Brevo. The author now wants to speak directly with 15 to 20 target users and give them a reason to pay one at a time. Cash costs and exact weekly hours are not disclosed.

First move: send plain-text issues to a small list, then ask target readers directly what they would pay for. A subscriber is a lead; this report contains no evidence that a subscriber became revenue.

3. Note AI side hustle: follower growth without sales

The week-seven report covers an AI side hustle on note, with Kindle publishing and note operations among its stated keywords. The author had 28 followers, up three from the previous week, and 20 total articles. The weekly figures were 72 views, 10 likes, and ¥0 in sales. Views fell from 81 and likes from 28, even as the follower count rose. 4

This is a low-cash test only in the sense that the source gives no cash number. It does not disclose tools, hours, or the first step, so a beginner cannot reproduce the input from this report alone. The useful signal is the separation between audience movement and payment: 28 followers did not produce a paid transaction in week seven.

First move: treat follower growth, views, likes, and sales as four separate counters. Do not let the first three stand in for the fourth.

4. AI website-content audit: a service with a real but incomplete number

The author describes an AI-assisted service that audits website content and proposes improvements. The public preview reports ¥43,280 in sales over the latest 30 days, reached within 45 days. It also says the work fits into one hour in the evening and began with Claude-assisted product-page copy. 1

That is the strongest payout in this week's set, but the number needs a narrow label. It is sales, not stated net profit. The article is paywalled, so the public material does not disclose tool fees, customer count, acquisition cost, or a full method. The author also describes a longer AI-work background, which means "45 days" should be read as the age of this offer, not necessarily the age of the person's entire side-hustle skill base.

First move: start with one concrete deliverable, such as a product-page copy audit, then price the result rather than selling generic AI help. The source does not disclose the first customer's path, so the 45-day number is evidence of a reported outcome, not a repeatable acquisition recipe.

5. DevMarathon: the renewal is the result

DevMarathon has a developer side and a B2B hiring side. The author says the first paying subscriber joined about a month earlier on a $24 monthly plan. On the day of the report, that subscription renewed and another $24 payment arrived. 6

The source gives no build cost, time commitment, acquisition channel, or first step. It also makes the right claim for the evidence: one renewal is a stronger product signal than one first sale, but it is not product-market fit or a growth story. For a full-time employee, the lesson is narrow and practical: a paid month that renews answers a different question from a signup or a launch-day spike.

First move: get one person to pay for a defined monthly outcome, then watch whether the same person renews. Keep the $24 payment separate from any larger revenue claim; this report only supports the one subscription and its renewal.

6. Brief Work: a working marketplace with no demand test

Brief Work is a two-sided marketplace for short expert gigs: one deliverable, under $5,000, completed in under two weeks. Clients pay $50 to post a brief while experts apply free. Alan Halley says he built it solo and went live in May. Three months later, production data showed five registered users, four briefs, all four posted by him, and $0 in revenue from strangers. He paid a freelancer $50 for one real transaction, but he was the client. 5

The product was not the only missing piece. Acquisition consisted of a launch post, a couple of social posts, and an email to three friends. The author says he did no outbound and never personally asked anyone to post a brief. He frames the failure plainly: a marketplace needs a way to reach both sides before more features can help. Build costs and weekly hours are not disclosed.

First move: name the first ten people who could pay and the path to reach them before building the marketplace. This is the week's clearest high-barrier flop because the software existed, while the demand test did not.

What a 9-to-5 reader can act on this week

Start with a bounded input. The cleanest positive test used five hours a month and changed how those hours were allocated. Set the time ceiling before choosing the platform.
Sell a narrow service before expanding the offer. The website-audit report started from a specific product-page task and produced a reported 30-day sales number. Treat the result as self-reported sales, not profit, and price in tool costs before copying it.
Ask for payment before polishing distribution-heavy software. GitPulse found direct outreach more useful than cold community posts. Brief Work found that a working product could still have zero outside revenue when the founder never asked a buyer to use it.
Keep four counters separate: followers, views, users, and paying customers. This week's note report shows why the distinction matters: the first number rose while revenue stayed at ¥0.

The most actionable pattern is small and measurable input. The most useful flop pattern is also concrete: attention, registrations, and a working build can all arrive before payment. None of these reports makes a promise about next month's income, but each leaves a reader with a number to test and a missing step to look for.

참고 출처

1
AI website-content audit side-hustle reportnote.com
2
Rakuten ROOM report: from posting to data reviewnote.com
3
97 days building an automated GitHub idea newsletterindiehackers.com
4
AI side hustle, Week 7: ¥0 revenuenote.com
5
I built a two-sided marketplace and never once asked anyone to use itindiehackers.com
6
A tiny retention signal that meant more than my first saleindiehackers.com
7
97 days building an automated GitHub idea newsletterindiehackers.com