Scope a prototype that answers one question
Most prototypes fail because they try to show everything. A short framework for cutting a feature idea down to the one flow that tells you whether to build it.
A prototype is a question, not a product
Modern tools make it quick to build something that looks real. That’s the trap: it’s now just as easy to build too much. A good prototype exists to answer one question you can’t answer any other way, then it has done its job.
Step 1: Write the question
- Weak: “customers like it”.
- Better: “five of eight customers complete the new setup without help”.
- Better: “our sales team can demo it and two prospects ask when it ships”.
Finish this sentence: “We’ll build this properly if we see…”. Good answers are specific and observable.
Step 2: Pick the one flow
List every screen you imagine, then circle the shortest path that lets someone experience the idea and lets you see the answer. Everything else can be a static image, a placeholder or simply missing. Sign-up, settings and edge cases almost never belong in a first prototype.
Step 3: Decide what can be fake
- Data: realistic sample data usually beats a live connection.
- Logic: a person behind the scenes can stand in for an algorithm you haven’t built.
- Design: use your existing look so feedback is about the idea, not the colours.
Step 4: Plan who sees it
Name the five to ten people who will try it and how you’ll watch them. Write down, before you start, what result would make you stop. Then build, show, learn, and either hand it to engineering with a clear brief or move on with the sprint you saved.