Validation is not about getting people to agree with us; it is about reducing uncertainty so we can make better decisions.

What does it mean to validate an idea?

Validating a business idea means testing the fundamental hypotheses underlying a proposal through experiments and observable evidence, to determine whether there is sufficient support to continue, modify, or reject specific business decisions. Validation reduces uncertainty; it does not provide absolute certainty or make the entire idea simply true or false. When designing a business, we make claims: “Customers have this problem,” “They will pay this price,” “We can deliver the service,” and “The cost per sale will be acceptable.” Some are supported by observations; others are assumptions. Experimentation means seeking relevant evidence to distinguish between them.

Strategyzer describes a Design–Test Loop: designing business alternatives and testing the hypotheses that support them. The results may lead us to stay the course, change important elements, or stop the initiative. The purpose is not to accumulate experiments but to reduce the risks that matter most.

The word “validated” can create a false sense of certainty. A test provides evidence under specific conditions; it does not prove that something will always work, in every market and at every scale.

To avoid confusing a fact with an explanation, distinguish observations, assumptions, hypotheses, and evidence.

1. Turn assumptions into testable hypotheses

A useful hypothesis states a specific claim about a behavior or condition. “Our application will be successful” is too broad. “At least 10 of the 30 restaurants contacted will agree to take part in an operational trial for two weeks” is more observable, although the chosen threshold needs a justification grounded in the context.

Classify hypotheses by risk type: desirability—do people want it?—feasibility—can we deliver it?—and viability—do the economics work? A fourth dimension can consider adaptation to the context and changes in it.

The decisive question is: What would have to be true for this business to work? Write it down before choosing tools or building prototypes.

Explore the types of business hypotheses and learn to write testable hypotheses.

2. Prioritize what is important and uncertain

Not all hypotheses deserve the same effort. An assumption is critical when, if false, it threatens the model or requires major changes. Its priority increases if we also lack solid evidence.

For example, before perfecting the visual design of a delivery service, we may need to establish whether recurring demand exists and whether delivery costs allow a reasonable margin. Testing secondary details first can create activity without reducing the main risk.

If the critical unknown is economic, review how to calculate costs, margins, and break-even to define which contribution assumption needs testing. A calculation based on assumptions does not establish that customers will buy or that real costs will match.

A simple log includes the hypothesis, its importance, current evidence, the proposed experiment, the person responsible, and the pending decision.

Prioritizing hypotheses by impact and uncertainty helps determine what to investigate first.

3. Choose the experiment to fit the question

An interview helps us understand experiences and motivations, but it does not prove on its own that someone will buy. An informational page can measure interest in a proposition, but a click is not a sale. A presale or paid pilot can provide behavioral evidence closer to an actual decision, although it also creates obligations we must fulfill.

Choose the least expensive test that can produce evidence strong enough for the decision at hand. Not every idea needs a digital MVP. A service can be tested manually; a physical product through samples; an operation through a controlled pilot.

Strategyzer emphasizes the importance of relevant participants, precise hypotheses, and appropriate artifacts. A test with people outside the target market can produce misleading results.

To compare methods by time, cost, and evidence, see how to choose an experiment. If testing calls for an artifact, review which prototype fits the question.

4. Define what you will observe before running the test

To avoid reinterpreting every outcome as a success, document the objective, population, procedure, metric, decision criterion, duration, and limitations before the test. The threshold must be justified by the business economics or the risk, rather than chosen because it “sounds good.”

Suppose we are testing a lunch offering. We can observe how many people actually pay, how many order again, and how much it costs to fulfill each order. If we measure only how many people say they like the menu, we could overestimate demand.

Avoid strong statistical conclusions when the sample is small or biased. The aim of an exploratory test may be to identify problems and generate hypotheses, rather than to estimate the behavior of an entire market precisely.

Population, procedure, and metrics are covered in how to design an experiment; bias controls should be planned before results are collected.

5. Run the test without manipulating the outcome

Document what you actually did, whom you involved, what changed during the test, and what limitations arose. If you change the price, audience, and message simultaneously, it will be difficult to know what produced the result.

Responsible experimentation also requires transparency. Do not charge for something you cannot deliver without explaining the conditions. Respect personal data, consent, and relevant legal obligations. Learning does not justify deceiving customers.

If a test involves personal data, bookings, advance payments, or a regulated activity, see how to investigate and document legal and operational requirements for the actual service and jurisdiction. Ethical experimentation does not replace checking specific applicable obligations.

An objective results documentation guide helps preserve events and incidents without later reinterpretation.

6. Interpret the evidence and decide

Separate three levels: data (“12 people paid”), interpretation (“The offering seems to attract this group”), and decision (“We will repeat the pilot through another channel”). Data does not speak for itself; we must consider participant selection, context, costs, and alternative explanations.

Possible decisions include persevering, adjusting an element, changing a central hypothesis, repeating a better-designed test, or abandoning the idea. A negative result can be valuable if it prevents a larger investment.

You can explore experiment data analysis and represent your options with a decision tree.

7. Return to the point that needs revision

The cycle does not require us to return to the same step every time. If we discover that the problem was different, we return to exploring the opportunity. If the price does not work, we revisit the model. If the prototype cannot be produced at an acceptable cost, we may need to redesign the solution and its preparation. Iteration moves across the first five pillars according to what we learn.

When evidence requires changing segments, channels, costs, or revenue mechanisms, update the Business Model Canvas and its assumptions before designing the next test. Revisiting the model does not mean the new choices have already been validated.

To apply these lessons in successive cycles, see iteration through the Design–Test Loop.

An example: will restaurants pay to manage reservations?

First, you interview restaurant managers and observe their current methods. You discover that mistakes occur mainly at certain times. You design a minimal service and test it with a few restaurants willing to use it. You record time saved, errors, onboarding effort, and willingness to keep paying. If there are savings but the cost of using the service is too high, the next decision may be to simplify the workflow before investing in full development.

What you should have by the end

A hypothesis log, documented experiments, evidence with explicit limitations, and justified decisions. You do not need to declare an idea “validated” as a definitive status. You need to know which relevant uncertainties you have reduced and which still prevent you from moving forward responsibly.

Key sources: Strategyzer [F2–F4, F7], IDEO [F1].

Bibliography and references

Numerical examples are illustrative and do not represent market data. Legal requirements depend on the jurisdiction and industry.