Prototype: Making Ideas Tangible
A prototype is anything that makes an idea tangible enough for people to react to: a sketch, a cardboard model, a clickable mock-up or a role-played service. Prototyping lets you learn cheaply before investing heavily, because it is far easier to change a drawing than finished software.
Why prototype?
- To learn: users react honestly to something they can see and touch, far more than to a description.
- To communicate: a prototype aligns your team and helps you explain the idea to partners.
- To fail cheaply: discovering a flaw in a paper prototype costs minutes; discovering it after launch costs months.
- To decide: comparing two prototypes helps you choose between ideas with evidence.
Fidelity: from rough to realistic
Fidelity describes how close a prototype is to the finished product. Low-fidelity prototypes are quick and cheap; high-fidelity ones look and work like the real thing but cost much more time.
Types of prototypes
| Type | Description | Good for testing |
|---|---|---|
| Sketches and storyboards | Drawings of screens or a comic strip showing how someone uses the solution | Whether the overall idea makes sense |
| Paper prototypes | Hand-drawn app screens on cards; you swap cards as the user "taps" | Navigation, wording, layout |
| Digital mock-ups | Clickable screens made with a design tool or even slides with links | Look and feel, flow on a real phone |
| Physical models | Cardboard, clay or 3D-printed versions of a device | Size, handling, comfort |
| Service prototypes and role-play | Acting out a service with real people (a delivery, a clinic visit) | Interactions, timing, emotions |
| Wizard of Oz | The front looks automatic, but a person does the work behind the scenes | Whether people value the result before you automate it |
How to prototype well
- Decide what you want to learn. Write one question per prototype: "Will vendors record a debt in under 10 seconds?"
- Build only what is needed to answer it. If you are testing wording, you do not need working payments.
- Make it quickly. Set a time limit, such as two hours.
- Make several versions so users can compare and you are less attached to one.
- Plan the test before you build (see Test: Learning from Users).
Otieno, 23, in Kisumu, wanted to help boda boda riders save for motorbike repairs. Instead of building an app, he drew five phone screens on index cards: a balance, a "save KES 50 today" button, a goal bar and a reminder. He sat with eight riders at a stage and asked each to "use" the app while he swapped cards. Riders ignored the goal bar but loved the reminder, and several asked whether savings could be locked until a repair was needed. Next, he ran a Wizard of Oz pilot: riders sent savings to an M-Pesa number and Otieno updated balances by hand in a spreadsheet and sent SMS confirmations. Within three weeks he knew which features mattered, without writing any code.
Making the first prototype too polished. Users are reluctant to criticise something that looks finished, and you become emotionally attached to it. Keep early prototypes rough enough that changing them feels easy. When handling real money in a Wizard of Oz test, keep careful records and be transparent with users about how the service works.
Key points
- Prototypes make ideas tangible so people can react and you can learn cheaply.
- Fidelity ranges from quick sketches to coded products; start low and increase only as you learn.
- Choose the prototype type that answers your specific question: paper, digital, physical, service or Wizard of Oz.
- Make several rough versions and plan how you will test them.
Tip: use the ← and → keys to move between pages.
