The most expensive thing you can build is the thing nobody wanted.
MVP development South Africa founders can learn from: a working prototype put in front of real people quickly, built to answer the one question the whole idea rests on rather than to be the finished product in miniature.
The cheapest question you will ever pay to answer.
What actually kills a new product
The best-known number in this field is that most startups fail because they build something the market did not need. It has been quoted for over a decade, and when the research was redone with several times more data it did not soften - poor product-market fit came out as the largest single cause of failure by a clear margin, ahead of timing and ahead of unit economics.
The same work makes a point that is easy to miss and matters enormously. Running out of money is recorded in the overwhelming majority of failures, and it is almost always the symptom rather than the cause. The money ran out because it was spent building the wrong thing carefully. A cheaper build would not have saved those products; a cheaper *question* would have.
And this is the part that should be encouraging rather than grim. "Nobody wanted it" is not a verdict you have to wait for the launch to receive. It is knowable beforehand, cheaply, by putting something in front of real people and watching what they do - and yet only a minority of founders formally test anything before committing. Most of the risk in a new product is concentrated in one or two assumptions, and both can usually be tested in weeks for a fraction of the build.
A prototype is not a small version of the product. It is an instrument for answering a specific question, and it is allowed to be crude everywhere the question does not live. That distinction is the whole discipline, and it is why a good prototype is usually finished before a founder expects and looks worse than they were hoping.
43%
Of failed startups died of poor product-market fit - the largest single cause when the research was rerun across 431 failures.
70%
Ran out of money, which reads as the cause and is the symptom. The capital went into building the wrong thing well.
40%
Of founders formally validate anything before launching. The cheapest step in the whole process is the one most often skipped.
The discipline
A prototype answers one question. These four decide which
Work through them in order and the scope writes itself. Skip them and you build a small version of everything, which answers nothing and costs more than it should.
The risk
What would sink this
- The assumption everything rests on
- What you believe about the customer
- What you believe about the money
- Which of these is least proven
This holds it up
The smallest test
The cheapest honest answer
- What must be real
- What can be faked convincingly
- What is out entirely
- How quickly it can be in front of someone
This holds it up
The build
Real enough to be used
- A working flow, not a slide deck
- Convincing where it is seen
- Fast to change
- Instrumented from the start
They see this
The verdict
You decide with evidence
- What people did, not what they said
- Where they stopped
- Whether anyone tried to pay
- Build, change, or stop
They see this
The verdict has to be allowed to be no. A prototype that cannot fail is not a test, it is a demonstration - and a demonstration is what you commission when you have already decided, which is a fine thing to want but a different and more expensive product.
What arrives
Everything on this list is yours at the end of it
Not a summary of what we will think about. The actual files, documents and decisions that land on your desk, stage by stage, and every one of them is yours to take anywhere afterwards.
01
The risk
- The assumption everything rests on
- What you believe about the customer
- What you believe about the money
- Which of these is least proven
A single named question the prototype exists to answer, agreed in writing, with the criteria for what counts as a yes and what counts as a no.
You walk away with
- The single question the prototype exists to answer, agreed in writing
- What counts as a yes and what counts as a no, decided before you see any results
- Every assumption listed and ranked by what it would cost you to be wrong about it
02
The smallest test
- What must be real
- What can be faked convincingly
- What is out entirely
- How quickly it can be in front of someone
A scope that is deliberately crude everywhere the question does not live - which is usually most of it, and is what makes the whole thing affordable.
You walk away with
- A scope that says plainly what is real, what is faked and what is out entirely
- A fixed timeline, usually weeks rather than months
- The plan for who will actually use it, and how they are recruited
03
The build
- A working flow, not a slide deck
- Convincing where it is seen
- Fast to change
- Instrumented from the start
Something real people can actually use, built to be thrown away or rewritten, with the behaviour recorded so the answer is evidence rather than opinion.
You walk away with
- A working flow real people can use, not a clickable slideshow
- Instrumented, so what happened is recorded rather than remembered
- Built to be changed or thrown away, which is what keeps it fast and cheap
04
The verdict
- What people did, not what they said
- Where they stopped
- Whether anyone tried to pay
- Build, change, or stop
A written answer to the question you started with, and a recommendation - including the one nobody wants, which is that this version should not be built.
You walk away with
- A written answer to the question you started with
- Where people stopped, named by step rather than described in general
- A recommendation: proceed, change, or stop
- A scoped path to the real build if the answer is yes
Measurement
How you will know it is working
MVP development South Africa is measured on whether it answered the question, not on what got built. These are the signals we watch.
What we report on
- Whether people completed the core flow without being helped through it
- Where they stopped, named by step
- Whether anybody came back unprompted
- Whether anybody tried to pay, or asked when they could
- A clear verdict on the original question - proceed, change, or stop
Get prototypes scoped for your business
Tell us what is happening now and you get a written scope with a price on it, from the person who would do the work.
If we think this is not actually your problem, we will say so on the call rather than after the invoice. It costs us the job often enough to be worth saying out loud.