Assumption-First Experiment Design
Test the riskiest business assumption before building
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 99%
Assumption-First Experiment Design starts by giving an idea a quick business-model shape, including its customers and economic logic. The team then asks what must be true for that model to work and lists the underlying hypotheses: whether customers have the problem, whether they will pay, whether costs are viable, and whether supply can work. Those assumptions are prioritized by risk rather than by the team's enthusiasm for the product. The highest-risk item receives the lightest credible experiment, such as customer discovery, card sorting, or an interest test. Evidence updates the model and determines the next test. Building is delayed until it becomes necessary to answer an important question. The mechanism is explicit uncertainty reduction, not executing a product plan as though its assumptions were facts.
Origin
Extracted from The Foundr Podcast
Core principles
- 01Ideas need a business-model shape before testing
- 02The riskiest assumption is often not about the product
- 03Experiments should reduce uncertainty before construction
How to run it
- 1
Shape the model
Spend a short, bounded period outlining the target customer, value proposition, revenue logic, and major costs.
Pro tip The transcript suggests an hour or two, accepting that the first version will be wrong.
Watch out Do not spend weeks polishing an untested canvas.
- 2
List what must be true
Write every important assumption underneath the proposed business model.
Pro tip Include customer need, willingness to pay, costs, and supply constraints.
- 3
Prioritize the assumptions
Rank the hypotheses by how uncertain and consequential they are if false.
Watch out The product assumption is not automatically the riskiest one.
- 4
Design the experiment
Choose the smallest credible test that can produce evidence about the top-ranked assumption.
Pro tip Consider interviews, card sorting, or an interest test before software or inventory.
- 5
Learn and reshape
Use the result to revise the business model, reprioritize assumptions, and select the next experiment.
Watch out Do not reinterpret weak evidence merely to protect the original idea.
- 6
Build when required
Construct only what is needed when a material assumption can no longer be tested credibly without it.
Watch out Activity is not the same as uncertainty reduction.
In the wild
Illustrative example: before building a scheduling tool, a founder sketches the customer and pricing model, then identifies willingness to pay as the highest-risk assumption. They conduct structured interviews and test interest in a paid offer before commissioning the full product.
→ The founder obtains pricing evidence before incurring most development cost.
Common mistakes
Starting with build
The build-measure-learn phrase can pull teams toward construction even when the main question can be tested without a product.
Testing low-risk details first
Easy tests create activity while leaving the assumptions capable of killing the business untouched.
Leaving the model unshaped
Without a quick model, the team cannot see the full set of desirability, viability, and cost assumptions it needs to test.
Is it for you?
Best for
It is best for founders moving from an idea toward a testable business model.
Not ideal for
It is not ideal when a minimal build is itself the cheapest credible way to test the highest-risk assumption.
From the transcript
“what needs to be true for this idea to work”
“come up with all 32 00 of the assumptions all of them underlying your your business idea your business model then you prioritize”
“it's one of the biggest mistakes that people think they need to build something to test”
From the episode
338: Foundr BEST OF 2020