Every product in the CoolSoftware workshop ships with an export button before it ships with a pricing page. That’s not a courtesy. It’s the whole design philosophy, working backwards from a single question: if I disappeared tomorrow, could you still get your stuff out?
Lock-in is a bug, not a moat
The conventional startup wisdom says stickiness is good — make leaving painful and retention takes care of itself. I’ve watched that strategy rot enough products that I no longer believe it. Lock-in doesn’t make a tool better; it makes switching away from a worse tool harder. Those aren’t the same thing, and users can feel the difference.
Anti-fragile tools get stronger when the user has options. If you stay because leaving is easy and you still don’t want to, that’s real retention. Everything else is a hostage situation with a subscription.
What “a way out” actually means
Concretely, for me it means three things:
- Your data is yours in a real format. Not a proprietary blob — JSON, CSV, Markdown, something a human or another tool can read.
- Self-hosting is a first-class path, not a footnote. If the hosted version goes away, the software still runs on your own box.
- No silent format drift. Exports stay stable, so a backup you took a year ago still imports today.
The selfish reason
There’s a self-interested angle too: building for portability forces cleaner architecture. If the data has to leave the building intact, you can’t let it tangle itself into your UI state and your database schema and three caches. The export button is a constant pressure toward keeping the core honest.
Twenty-five years of maintaining other people’s lock-in taught me to hate it. Now every tool I make is built to be left — which, paradoxically, is the main reason people stay.