This is a placeholder long read, used to test the table of contents and reading time for long pieces. Delete it once the real article is written.
The answer: longer than you estimate, shorter than you fear.
Week one: ideas are cheap
The idea stage is the most comfortable, because nothing has been exposed yet. I wrote down seven features on paper, and five of them never got built. The two that actually survived were the two I use myself every day.
The easiest way to decide whether a feature is worth building is to ask: If I don't build it this week, will I catch myself doing it by hand anyway? If yes, build it; if not, cut it.
Weeks two to four: the real work is in the corners
The main flow was working in three days. The remaining three weeks all went into:
- Empty states — what the page looks like the first time you open it and there's nothing there;
- Error states — the network drops, a field is filled in wrong, a permission isn't granted;
- Mobile — a layout that looks good on desktop falls apart completely at 375px wide;
- Copy — every single prompt had to be rewritten three times before it stopped sounding like a machine.
Not one of these four was in the original estimate.
Week five: reasons not to launch
By week five the product already worked, but I kept not shipping it. I could come up with a new reason every day: the icon isn't good enough, the docs aren't written, it's still missing a share feature. Later I realized none of these were reasons. It was fear.
The cure is pretty low-tech: set yourself a date, and tell one person.
Conclusion
For one person finishing something usable, four to six weeks is normal. Writing code takes up roughly 40% of that; the other 60% is tidying up edges, writing words, and arguing with yourself.
If you're in week five right now, ship it. Honestly, nobody is going to stare at that icon the way you do.