I built 12 apps in 10 months and made $10k. Here are 10 things I wish I knew when I started
Over the last ten months, StoicSoft — one person, one VPS — built twelve paid desktop apps: a terminal for AI coding agents (1DevTool), a server manager, a notes vault, a file tool, a marketing tool, a video tool and more. Total revenue just crossed $10k by the author's own account — about $8.6k after tax and refunds. His words: not a startup story, not a failure either, but the boring middle where most indie devs live.
The lessons he says he wishes someone had told him in March:
- "Build mode" is a comfortable way to avoid selling. The revenue split is lopsided: 1DevTool made $4.8k (56%), bundles $1.5k, Server Compass $1.3k — and four apps made $0. 1DevTool was the only app he ever marketed; zero sales, he concludes, meant never marketed, not bad product.
- Shipping releases is not marketing. He shipped 17 releases a month for three months while orders went 27, 23, 13 — then shipped nothing for two weeks on a trip, and orders went up.
- Every launch decays. Server Compass went from a $296 peak month to $59; 1AIVault faded to $0. Plan the second launch before the first.
- Your home audience runs out. 70% of 1DevTool buyers are Vietnamese; that revenue line decayed from $1,308 in August to $129 in the first half of November, while non-Vietnamese revenue grew. Go English earlier, he advises.
- Traffic to free things does not buy paid things. About 900 Google landings a day across his sites, a third going to a free 3D model library that sold exactly one discounted license.
- Packaging sold more than new apps. His best fortnight ($723) leaned on bundles and device upgrades; a $30 upgrade page built in a day outsold three of his apps.
- Label your checkout links from day one. Sales rose 1.8x in November with no release and he could not tell why — he only added source tags to buy links on November 15, eight months after the first sale.
- The 13th app is the most expensive. In July he built seven MVPs in three days with AI agents — 900+ tests, all green, none launched.
- Nobody asked for most of his 343 features. The most upvoted feature request has two votes; what people actually pay for is agents handing off work to each other — one feature buried under 342 others.
- Do the math before selling your best app. A buyer hinted at "$50k in range" for 1DevTool; at 1.5x–3x its roughly $8–9k yearly revenue, a realistic price was $12k–$25k, plus a 36-month non-compete. He did not sell.
His plan now, in his own telling: keep all 12 apps for two more years, give marketing a fixed share of every week, test one small app at a time with the pass mark written down in advance — and no 13th app.
1VR Pick · Source: Medium @StoicSoft
机器翻译 ·