Prototypes optimize for discovery. Products optimize for repeated use. The transition between them is not a single rewrite; it is a series of small decisions that remove uncertainty.
Add the missing edges
The center of a prototype usually works. The edges are where practice begins:
- an empty collection;
- an expired session;
- a long title;
- a slow connection;
- a second browser tab;
- a user who presses the button twice.
Write these cases down before adding another feature.
A practical sequence
| Stage | Question | Evidence |
|---|---|---|
| Observe | What do people actually do? | Session notes |
| Clarify | Where do they hesitate? | Support questions |
| Stabilize | Which failures repeat? | Error reports |
| Simplify | What can disappear? | Before/after flow |
Make the result inspectable
A good tool leaves useful traces:
{
"action": "publish",
"status": "completed",
"durationMs": 842
}The trace should help answer what happened without exposing secrets or forcing someone to reconstruct the entire interaction.
Finish with a boring checklist
- Empty state has a clear next action.
- Long content wraps without breaking the layout.
- Errors explain how to recover.
- The destructive action asks for confirmation.
- The help link points to a maintained page.
The last two unchecked boxes are not failures. They are visible work, which is much safer than invisible assumptions.