Experiment
Also called: Assumption test
Last reviewed
An experiment is a small, fast test that checks a specific assumption behind a solution before the team commits to building it. In an Opportunity Solution Tree, experiments hang below solutions and produce the evidence that decides whether a solution advances, changes shape, or gets dropped.
Test the assumption, not the solution
You rarely test a whole solution at once. You pick the assumption that would sink it fastest if it turned out to be false, and you design the cheapest test that would tell you. A solution with five assumptions might need only one experiment, if one of those assumptions is clearly the riskiest.
Choosing what to test first
Rank assumptions by two questions: how damaging is it if this is wrong, and how much evidence do we already have? The assumption that is both high-risk and poorly evidenced is where the first experiment goes. Everything else can wait.
Deciding in advance
An experiment without a pre-agreed threshold tends to confirm whatever the team hoped for. Writing down what result would count as a pass, and what would count as a reason to stop, before running the test is what turns an experiment into a decision rather than a data-gathering exercise.
Frequently asked questions
What is the difference between an experiment and an A/B test?
An A/B test is one kind of experiment, and usually a late one, because it needs a working implementation and real traffic. Most discovery experiments happen earlier and cheaper: interviews, prototypes, fake doors, concierge tests and data analysis all count.
How long should a discovery experiment take?
Days rather than weeks. The value of an experiment comes from changing a decision, so a test that takes longer than the decision can wait for has already failed at its job, regardless of what it finds.
Related terms
- Assumption testingAssumption testing is the practice of identifying the beliefs a solution depends on and checking the riskiest ones with small tests before building.
- SolutionA solution is a specific thing a team could build or change to address one opportunity in an Opportunity Solution Tree.
- EvidenceEvidence is the recorded observation that supports or undermines a claim in an Opportunity Solution Tree.
See how experiment fits the whole tree
The full guide covers all four layers and how they connect.