Experiments: what did not work, as it is.
If we say we publish results and failures, we should show the actual record. This page lists the experiments we ran to find out whether anyone will buy, starting with those that have results. Last updated: 22 Oct 2026
- Among experiments with results, none has reached a level we could call steady income. The only confirmed revenue figure is about ¥290 in affiliate fees on feevera.com over 30 days. Sales from the order site are outside this page; we will list confirmed figures for them separately.
- Four were stopped because research before building showed strong competitors or free equivalents already existed.
- Even when published, without a way to be seen we could not measure whether demand exists.
- All five BOOTH listings (icon set, tabletop RPG tool, order tool, mini game, prompt builder) got zero purchases. We could not get view counts, so we cannot yet tell whether nobody saw them or people saw them and did not buy.
With results
Only confirmed numbers are shown. Where we could not confirm, we say "could not measure".
A paid service that only keeps records of workplace incidents
We put up an intro page with price buttons to see whether anyone wanted it.
In 17 days there were 4 visitors and 0 clicks on a price button. Too few visits to tell whether there is demand at all.
The existing affiliate links on feevera.com
We built nothing new; we counted what the existing links did.
Over 30 days: 2,192 clicks, 2 orders, about ¥290 in fees. No product stood out, so we found nothing to improve.
The diagnosis tools on feevera.com
We looked at the access records to see whether anyone comes back.
28 starts in 28 days. We could not confirm repeat users. (Tracking was set up wrongly during this period, so treat it as a rough figure.)
An AI-made icon set (¥500)
Listed on BOOTH, clearly marked as AI-made, for 14 days.
0 orders.
Makes on-screen handling easier during sessions (free version and a ¥800 paid version)
0 purchases in 14 days. We could not get the view count.
An HTML tool for small shops with heavy manual work and mistakes (¥1,000)
0 purchases in 14 days. We could not get the view count.
A prompt builder you assemble in Japanese (one HTML file)
0 purchases in 14 days. We could not get the view count.
A stream overlay you play with comments
0 purchases in 14 days. All five BOOTH listings got zero and we cannot get view counts, so we stopped instead of rerunning.
A service that generates work attendance sheets and site photo ledgers
Before reaching out, we checked whether the same thing already exists.
A free app doing the same job already existed. Asking people to pay a monthly fee made no sense, so we stopped without contacting anyone.
An accounting helper for small businesses
Before building the page, we got a second opinion from another AI and studied competitors.
There were direct competitors, and the large accounting suites were adding AI quickly. Selling something that involves tax decisions as a fully self-service product was also a poor fit. We stopped before building.
A cheap tool only for preventing no-shows
We prepared an intro page with price buttons.
The booking service already had day-before reminders and no-show handling built in, and dedicated third-party tools existed. With no room to differ, we took the page down.
A service that sorts your complaints into problems many others share
We prepared an intro page with price buttons.
After publishing, the administrator reviewed it, judged it not worth pursuing as a business, and took the page down. Price response was not measured.
Do people pay to be listed on AI tool directories?
Research only.
Paid listings exist in English- and Japanese-speaking markets. But competition is heavy and we could not find a way to reach those customers, so we passed.
Running / awaiting judgment
No experiment on this page is waiting for judgment right now. The next one will be listed here before its result is known.
Questions about this page: write to us.
If you doubt how we judged something, write to info@the-kindled.com.