OneManSaas

27 years as an software engineer. I have a son with polymicrogyria who needs more of my time than a 9-5 allows. Building SaaS products in public to replace my salary and be there for him. Also running a $120k/year eBay reselling business and managing rental properties. Building starts now.

A refund isn't an undo. PayPal keeps the flat per-sale fee even when you hand the money back, so a cancelled sale leaves you worse off than no sale. The smaller your average order, the bigger that bite. Treat refunds as a cost line, not a reversal. paypal.com/us/legalhub/paypa…

The glass on the first iPhone was invented in 1962. Corning found no carmaker willing to pay extra and shelved it by 1971. Apple called in 2007. "Nobody wanted it" describes a moment, not the work. Keep a list of what you shelved and why you shelved it. forbes.com/sites/ewanspence/…

L.L. Bean's first product failed. 90 of the first 100 pairs fell apart. He refunded every one, borrowed to fix the stitching, reshipped free. That famous guarantee was the bill for a bad launch. Price your own worst case before you sell anything. encyclopedia.com/social-scie…

Wrigley sold soap and threw in free baking powder. The baking powder outsold the soap. So he sold baking powder and threw in free gum — the gum outsold that too. He dropped the rest. If your freebie is what people come for, that's the product.

Most founders treat a patent as a wall. It's a countdown. Nestle had about 1,700 patents around Nespresso; the core capsule patent ran out in 2012 and compatible capsules flooded in. Use the protected years to build habit and brand. Those don't expire. vennershipley.com/insights-e…

Most founders think distribution is something you buy. Dollar Shave Club's launch video cost $4,500 and took a day to shoot — 12,000 orders in 48 hours, then a $1B sale to Unilever. Make one thing worth forwarding before you pay to push an ordinary one. inc.com/magazine/201707/lind…

People put in less effort when their own part can't be told apart from the group's. It's one of the more reliable findings in psychology: 78 studies, 15,000 people. The fix isn't motivation, it's giving each person a slice whose result shows on its own.

How do you estimate something you can't look up? Ask a few people separately, then take the middle answer. At a 1906 fair, 787 people guessed an ox's weight; the middle guess landed within 1%. Ask them together and you just get one guess, repeated. nature.com/articles/075450a0

Your API is not a moat. In 2021 the US Supreme Court ruled 6-2 that Google copying 11,500 lines of Java's interface was fair use. Put the moat in your data and accounts - and treat rebuilding an incumbent's interface as a way in, not a legal wall. copyright.gov/fair-use/summa…

A test can be huge, clean, and still point the wrong way. Coca-Cola ran ~190,000 blind taste tests and people picked New Coke. It was pulled 79 days after launch. Nobody was asked how they'd feel losing the old one. Test what customers stand to lose, not just what they prefer.

Rewrite it from scratch? Netscape did. Its last big release was June 1997; the rewrite shipped November 2000 - three years, five months of silence while Internet Explorer took the market. Price a rewrite in months your rivals get free, not in dev hours. en.wikipedia.org/wiki/Netsca…

A big review isn't a careful one - it's an approval with extra steps. Across 2,500 code reviews, people caught far fewer problems past about 400 lines at a sitting, and the most under 200. Attention runs out before the change does. Send work in smaller pieces.

Why do ads on your own brand name always look profitable? They're shown to people already coming to you. eBay switched its brand ads off in a controlled test and traffic barely moved. Switch yours off for two weeks and watch orders, not clicks. nber.org/papers/w20171

Writing publicly about how your product works is a legal act, not just marketing. A dated post stops anyone else from patenting the idea. It also ends your own patent option in Europe the day it goes up, and starts a 12-month clock in the US. File before you publish.

You don't get cheaper by waiting. You get cheaper by doing it more times. Solar panel prices fell about 20% every time the world's installed total doubled - for four decades. Count reps, not months: doubling your onboardings is what cuts their cost. ourworldindata.org/learning-…

A lot of rules founders blame on their payment processor are actually federal law. In the US you can refuse credit cards under $10, but you cannot set any minimum on debit. If small orders are eating your margin, that credit minimum is a lever you already have.

Fixing a problem well doesn't make a customer more loyal than one who never had a problem. Pooled studies find it lifts satisfaction scores but doesn't move repeat purchases. So track whether your recovered customers actually buy again.

Why do some products feel more valuable when they're slower? In experiments, people preferred a travel site that showed its work to one that answered instantly - same results. Visible effort reads as effort spent on you. Show the work, not a spinner. hbs.edu/faculty/Pages/item.a…

A backup you've never restored is a hope, not a plan. In a survey of 1,000 tech executives, 62% said their org doesn't regularly test restoring from backups — most find out at the worst moment. Put one test restore on the calendar this quarter. cockroachlabs.com/blog/the-s…

Some trust badges aren't for sale at any price. Gmail's blue check next to a sender is issued against a registered trademark, not a payment. If you want one, the first step is a trademark filing - and that takes months, so start there.