August 3–10: the $578 utility, ¥6,130 eBay profit, and a £3 Etsy margin (2026)

The strongest low-cash result this week came from a tiny image-resizing utility: $17 in month one, then $578 in month two. The more sobering numbers sit beside it: an Etsy print-on-demand order worth £80 left roughly £3 profit, and a fact-checking browser extension was still at $0 after eight months.

Coverage runs from August 3, 2026 at 09:00 through August 10, 2026 at 09:00 in the channel's display timezone (UTC-05:00). Every payout below is self-reported, not independently audited. A sale, profit, revenue total, and asking price are kept separate.

Quick scan: ranked by barrier, then payout

This order favors a test a full-time employee could start without a large cash bet. Within a similar barrier, the larger reported result ranks first. Existing inventory and prior skill count as barriers even when the current week's cash outlay is zero.

Rank	Experiment	Reported result	Startup cash	Barrier and first move
1	Ad-supported image-resizing utility	$17 in month one, then $578 in month two	Started on a free site; later hosting cost not disclosed	Low-medium if you can make a small web tool. Built a script in a few hours and shared it in seller groups. 1
2	eBay watch parts and accessories	July sales ¥17,757, profit ¥6,130 from 4 sales	No new sourcing during the reported three months; original inventory cost not stated	Low cash, medium product knowledge. Kept existing repeatable parts listed while pausing sourcing. 2
3	Etsy framed artwork via Gelato print-on-demand	£80 first sale, about £3 profit after costs	£51.41 print charge, nearly £10 Etsy fees, discount and listing costs on that order	Medium. Made new products, used a print-on-demand supplier, and took the first order. 3
4	Small iOS app portfolio built with Claude Code	$384 revenue in the latest month	Not disclosed	Medium-high. Started developing iOS apps less than two months earlier and shipped several small apps. 4
5	Claude Code, paid Note posts and a posture app	¥1,500 cumulative sales in four months; fixed costs about ¥3,800/month	¥3,800/month after moving from Claude Max to Pro	Medium-high. Started during parental leave, published paid Note articles, released a free beta and built toward a ¥1,980 one-time app price. 5
6	TruthScore Chrome extension and web fact checker	$0 revenue after eight months; 3 users and 27 email subscribers	Gemini API costs money each month; amount not disclosed	High. Built alone, added a Chrome extension and web search, but still has no repeatable acquisition channel. 6

The table has one important trap: zero new cash is not the same as zero barrier. The eBay result depends on inventory already bought and a product line that can keep selling. The ad tool started cheaply, but its owner later reported that the host blocked it after more than 175 GB of data use. 1

1. The tiny utility that hit a hosting problem before it hit a revenue ceiling

The experiment was a simple web script for resizing product images. A seller uploaded an image, the script resized it, and Google AdSense appeared on the result page. The maker says the first month produced $17. After sharing the link in a couple of seller groups where people were already complaining about image sizing, the second month reached $578. 1

The first move was clear and small: solve one annoying file-format problem, put the tool on a free site, then take it to communities where the complaint already existed. No inventory or customer support was reported. The post does not give the first-dollar date, the traffic count, or the original hosting bill, so the revenue cannot be converted into a stable forecast.

The catch arrived after three months. The hosting partner blocked the site after it used more than 175 GB of data, and the maker planned to move to a VPS and add features. That turns the experiment into a useful warning for anyone copying the model: AdSense revenue can rise with usage while the delivery cost rises faster. 1

What a 9-to-5 reader can copy: pick a repetitive task with a visible complaint, ship the smallest useful transformation, and check the hosting or API cost before promoting it widely. The reported $578 is the upside signal; the 175 GB problem is the part that belongs in the starting budget.

2. eBay watch parts kept paying while the seller stopped sourcing

This Japanese eBay seller reported three months in which they bought no new inventory because software development took all of their available time. Existing listings still produced four sales in July: ¥17,757 in sales and ¥6,130 in profit. The three-month total was ¥57,551 in sales and ¥18,403 in profit across 13 sales. 7

The useful detail is the product mix. One-off watches disappear when sold. Repeatable watch bands and parts can stay listed and generate orders without another sourcing trip. The source describes July as four sales from these existing, continuously listed items, not as proof that a new seller can begin with no inventory. 7

That makes this a low-cash test for someone who already owns suitable stock, but a poor fit for a reader who would need to buy watches first. The report gives no separate time budget and no original acquisition cost. Its strongest evidence is about resilience during a busy month: a small catalog of repeatable items created a floor of roughly ¥5,000–¥7,000 monthly profit while sourcing stopped. 7

First test: list five items you already own, calculate profit after platform and shipping costs, and see whether any are repeatable rather than one-off. Do not treat another seller's existing stock as your startup capital.

3. An £80 Etsy order left almost nothing after print-on-demand costs

An Etsy seller restarted with new products and received a first £80 order for artwork in a wooden frame. The print-on-demand supplier Gelato charged £51.41. The seller also gave a 10% discount, paid nearly £10 in Etsy fees, and had listing costs. The seller estimated the final profit at about £3. 3

The attraction is obvious: no finished inventory had to be stored before the order. The margin is the real result. An £80 sale is not an £80 payout, and the seller's first new product became a near-break-even test after production, platform fees, discounting, and listings.

The source does not disclose a monthly revenue total, country, hours, or prior product-development cost. It is therefore a first-sale signal, not a monthly business case. The concrete first step was still useful: make new artwork, connect a print-on-demand service, and get one order into the real fee stack before producing a catalog.

First test: price from the landed cost backward. Include production, platform fees, discounts, listing charges and a refund buffer before deciding whether the product is worth making.

4. A beginner iOS portfolio reached $384, but the report skips the cost line

A maker reported starting iOS development less than two months earlier. Their apps generated $384 in the latest month, and most of the apps had been built about a month before the report. The maker says Claude Code handled much of the development and helped them ship faster. 4

The first step was not a single service offer: it was learning to build and release several small apps. That makes the experiment more scalable than a one-off job, but it also adds app-store distribution, product selection and maintenance risk. The report links to the developer's App Store page, but does not break out the apps, downloads, pricing, ad spend, developer fees or net profit. 4

So the $384 is a real self-report, not a clean beginner forecast. It shows that a new developer can reach paid distribution quickly; it does not show which app earned the money or whether the amount repeats.

First test: choose one narrow problem, release a small version, and track revenue by app. Do not let a portfolio total hide a product that earns nothing.

5. Four months of AI side work produced ¥1,500 and a decision to stop one line

A Japanese parent on parental leave reported ¥1,500 in cumulative sales after four months. The money came from two paid Note articles: a ¥500 article was bought twice in May and another ¥500 article once in early July. The creator also reported fixed costs of about ¥3,800 per month after switching from Claude Max to the cheaper Pro plan. 5

The portfolio included ongoing Note publishing, a free beta posture-monitoring app called ShiseiKeeper, and a planned one-time price of ¥1,980 for the app. The app had a working technical foundation but no payment button yet. A separate content activity stayed at zero revenue and was stopped in July because childcare left too little time to keep producing it weekly. 5

This is a better flop report than a simple "AI did not work." It gives a cumulative payout, a recurring cost, the exact paid product that generated sales, and a reason for cutting scope. The current result is still negative on the disclosed fixed-cost line: ¥1,500 in four months against roughly ¥3,800 per month in ongoing costs. That arithmetic covers only the reported fixed costs, not the value of the maker's time.

First test: if time is tight, keep the line that already has a sale and pause the one that only consumes weekly production time. A free beta is not a paid product until the checkout path works.

6. TruthScore has users, a working extension and still $0 revenue

TruthScore began after its maker was scammed by a make-money-online video. The solo founder built a tool that scores suspicious YouTube claims, then added a Chrome extension and web search so a user can check a video without leaving the page. The August 9 update reported 3 users, 27 email subscribers and $0 revenue after eight months of building and iteration. 6

The cost boundary is visible even though the amount is not: Gemini API usage costs the maker money every month, and there is no budget for paid ads. YouTube comments, the intended acquisition channel, were described as unusable after a shadowban and spam treatment. 6

The product is more developed than the revenue suggests. That is exactly why the report matters to a full-time employee: eight months of work, a working extension and real users still did not answer the commercial question. The bottleneck is distribution and payment conversion, not another missing feature.

First test: find ten reachable people with the problem before building the extension. If they use the free tool but do not pay, ask what would make the paid version worth buying before adding more infrastructure.

The practical read for this week

Need the cheapest proof? Start with an existing skill, existing inventory or a small utility. The image tool and eBay report both reached real payouts without a fresh inventory purchase, but each carried a hidden dependency: hosting capacity in one case, prior stock in the other. 1 7
Selling a physical product? Count the money left after production and platform fees. The Etsy seller's £80 order became about £3 profit, so the first sale validated demand more than it validated pricing. 3
Building software? Treat distribution and billing as part of version one. The iOS report has revenue but little cost detail; TruthScore has users but no revenue; both leave unanswered questions that more code alone will not solve. 4 6
Short on time? The ¥1,500 report is a reminder to keep the activity that has already produced a sale and stop the one that only creates a weekly production obligation. 5

The pattern is not "software wins" or "physical products lose." It is narrower: the low-barrier tests exposed their real costs early, while the hardest builds still had to find a buyer. That is the comparison worth carrying into next week's experiment.