Why you should prototype an app before you build it
Why a clickable prototype is the cheapest way to learn what works in an app, what to test with it, and what a prototype cannot tell you.
Changing a design is cheap. Changing working software is not. A prototype lets you make your mistakes while they still cost an afternoon instead of a month.
That is the whole argument, really. The rest of this article is about how to get the most out of one.
What is a prototype?
A prototype is a mock version of your app that looks and behaves enough like the real thing for someone to try it. There is no database behind it and nothing is actually saved. Tapping "Book now" simply jumps to a picture of the confirmation screen.
They come in a few levels of polish:
- Paper sketches. Screens drawn by hand. You swap sheets as the person "taps". Rough, fast and surprisingly effective for early ideas.
- Wireframe prototypes. Grey boxes and plain labels linked together so you can tap through on a phone. Good for testing structure and flow without getting distracted by colour.
- High fidelity prototypes. Finished looking screens with real text and images, linked into a clickable demo. Close enough to the real thing that testers often forget it is a mock up.
Design tools make the clickable kind quick to produce. A designer can usually link up the main screens of a small app in a day or two once the layouts exist.
What does it save you?
Think about what happens without one. You agree on a set of static screen designs. A developer spends weeks building them. You install the first version, hand it to a customer, and they cannot work out how to change their booking date. Now the fix involves redesign, rebuild, retest.
With a prototype, you find that problem in week two. The fix is dragging a button to a different spot.
Prototypes also settle arguments. Two people on your team disagree about whether the menu should be at the bottom or in a side drawer. You can debate it for an hour, or you can mock up both, hand each to a few customers, and watch which one they find easier.
What should you test with it?
Do not ask people whether they like it. Give them things to do and watch.
Say you are building an ordering app for a cafe. Useful test tasks would be:
- Order a large flat white with oat milk for pick up in fifteen minutes.
- Add a toasted sandwich to that order.
- You changed your mind about the sandwich. Remove it.
- Find out whether the cafe is open on Sunday.
- Reorder what you had last time.
While they work through these, look for:
- Where they hesitate. A pause of a few seconds usually means a label or layout is unclear.
- Where they tap first. If it is not where you expected, your screen is sending the wrong signal.
- What they say aloud. "I assume this means..." is a warning sign.
- Whether they finish. Completed, struggled, or gave up.
- What they expected to happen next. Ask before they tap: "What do you think this will do?"
Five people is enough to surface the big problems. Fix them in the prototype, then try again with a few more.
What can a prototype not tell you?
A prototype is a useful liar. It is worth being clear about what it hides.
- Performance. Everything in a prototype is instant. Real apps wait on networks. You will not learn how the design feels on a slow connection.
- Real data. Mockups have tidy names and perfect photos. Real life has a customer with a thirty letter surname, a product with no image, and a list with four hundred items.
- Technical feasibility. A designer can draw anything. Whether the booking system you rely on can actually supply live availability in that form is a separate question for a developer.
- Long term behaviour. A test shows whether someone can use the app once. It does not show whether they will open it again next week.
- Whether people will pay. Enthusiasm in a test session is not a sale.
None of this is a reason to skip prototyping. It is a reason to involve a developer early, so someone is checking the design against reality while you test it against people.
How polished should it be?
Match the polish to the question.
If you are asking "is this the right set of features, in the right order?", stay rough. Grey boxes are fine, and people give more honest feedback on something that looks unfinished. Show someone a glossy design and they comment on the colours.
If you are asking "is this clear and pleasant enough to release?", go high fidelity with real wording. Placeholder text hides problems, because the words on buttons and headings do a great deal of the work.
Before you approve any build
Ask whoever is designing your app for these:
- A clickable prototype you can open on your own phone.
- At least one round of testing with people who are not on the project.
- A short list of what changed because of that testing.
- A developer's review of anything that looks technically risky.
- Designs for the unglamorous states: empty lists, errors, no connection.
If testing changed nothing, be a little suspicious. Either the first design was unusually good, or the test was too gentle. In our experience it is nearly always the second.