Idea validation

Most new products fail because not enough people want them, not because they are badly built. This skill tests the beliefs behind an idea with small experiments before the founder spends months building.

When to use

• Before building anything, and again before each big investment of time or money.
• After customer-interviews, to turn what was heard into experiments.
• Not to judge an idea by opinion: the skill never says "good idea" or "bad idea" without data.

Inputs

1. The idea: target customer, problem, solution, how it makes money.
2. What is already known (interviews, data, competitors) and what is only believed.
3. Budget and time available for testing.

Method

1. List the assumptions in three families: desirability (customers want it), viability (it can make money), feasibility (it can be built and delivered). Write each one as a sentence that can be proven false: "Independent bakers will pay a monthly fee to take orders online."
2. Map them on two axes: importance (does the idea die if this is false?) and evidence (do we have proof?). Test first the important assumptions with the least evidence.
3. Design one experiment per top assumption, the cheapest one that gives a real signal:
   • Behavior beats opinion: a sign-up, a pre-order, a deposit or a letter of intent counts more than "I would buy it".
   • Options: problem interviews (customer-interviews), a landing page with a sign-up button (landing-page), a concierge test (deliver the service by hand to a few customers), a Wizard of Oz test (the product looks automated, people do the work), a pre-sale, a clickable prototype (prototype-testing).
4. Write the prediction before running the test, in Savoia's form: "At least X% of Y will Z" (for example, "at least 10% of 200 visitors from the bakers' forum will leave their email within 14 days"). The target comes from a cited benchmark or is labeled as the founder's own guess.
5. Run it, then decide with the criteria set in advance: persevere, change one element (pivot), or stop. Write down what was learned, even when the test fails.
6. Watch for false positives: friends and family, compliments, answers about the future, tiny samples, and traffic that is not the target.

Output

### Assumptions
| # | Assumption (can be proven false) | Family | Importance | Evidence today |
|---|---|---|---|---|

### Experiments, riskiest first
| Assumption | Experiment | Prediction (at least X% of Y will Z) | Cost | Duration |
|---|---|---|---|---|

### Decision rule
Persevere if … · Pivot if … · Stop if …

Quality bar

• Every assumption can be proven false.
• Each experiment measures behavior, not opinions, whenever possible.
• Predictions, samples and durations are set before the test; targets come from a cited source or are labeled as guesses.
• No verdict on the idea without evidence.
• Written in the user's language, with the terms below.

Terminology

Sources

• Ries, E. (2011). The Lean Startup. Crown Business.
• Bland, D. J., & Osterwalder, A. (2019). Testing Business Ideas. Wiley.
• Savoia, A. (2019). The Right It: Why So Many Ideas Fail and How to Make Sure Yours Succeed. HarperOne.