In short
A prototype is an early, testable version of a product or feature that lets you check ideas with users before time and budget go into development.
What is a prototype?
A prototype is a draft you can actually try out. It simulates how a product or feature should work without being fully built. That might be a paper sketch you "click" through with your finger, a clickable mockup in a design tool or a quickly coded version with real data.
Prototypes are usually grouped by fidelity. Low-fidelity prototypes are rough and quick to make, which is ideal for structure and flow. High-fidelity prototypes look almost like the finished product and are suited to testing details, content and interactions.
Why prototypes matter
Changing a prototype takes hours. Changing finished code often takes weeks. A prototype helps you find comprehension problems, missing steps and wrong assumptions before development begins. It also makes ideas tangible: stakeholders discuss something concrete instead of descriptions, and engineers know more precisely what to build.
How to use prototypes in practice
- Start with the question: Decide what the prototype needs to answer, then pick the right level of fidelity.
- Build only what you need: Cover the flows you want to test, not the whole product.
- Use realistic content: Real product names, prices and copy produce better test results than placeholders.
- Test with users: Even five participants often reveal the biggest problems.
- Be ready to throw it away: A prototype is a learning tool, not an early draft that has to survive.
Example
A direct-to-consumer furniture brand is planning a product configurator. Instead of sending it straight to development, the team builds a clickable prototype and tests it with five people. It turns out fabric selection should come before size. In the prototype, that change takes an hour instead of a sprint.

written by
Benedikt Geisberger
Got the theory? Let’s test your product.
We use real user tests to show you where your product gets stuck and what to improve first.