If users trust you with their data, you have an obligation to stay alive.
You are the custodian of hours of effort entering data into your system, they don’t want to be told that it’s “easy to export” it after you shut down your servers.
Footprints of giants…
A brief list of notable survivors from across the ecosystem.
SQLite: They try really hard to have zero dependencies.
NetNewsWire: The feature set is small, the functionality is client-side, and the app is open-source.
Things: No free tier, intense platform focus.
Bear: Platform focus and no operational costs (all server-side functionality is outsourced to iCloud).
Scapple: One-time license and no dev work for years. The “make it work the first time and don’t touch it again” theory of development.
Things you can try
We don’t have perfect answers here.
Don’t raise money.
It forces you to make money.
Ship slowly, and “finish” things
It lets you have a smaller team, and carefully considered feature surface.
Large markets require constant work on features and bug fixes, if you can arrive at a stable product that does what it needs to for its intended user base, you can take the foot off the gas and declare it “complete”. You now no longer have a “maintainer’s salaries” dependency.
Raise your bus factor.
How many deaths can your web app keep runnning through?
Minimise parts of the system, so a single credit card can keep them all running, etc.
Open source
Unfortunately this doesn’t make much sense for web apps/publishing tools, because most of the value is in having your stuff running in the cloud.
Export formats
Markdown is readable and editable. HTML is editable and publishable. JSON is parsable and more directly convertable. CSV is somewhat readable and editable.