At the early stages of product or business discovery, a company has not yet committed significant capital to a new business idea or product initiative. It is still trying to determine whether the idea “will work” and whether there is “a real opportunity” ahead. This is the moment when relying only on intuition becomes dangerous. Intuition and domain expertise are essential for generating original business ideas. But when it comes to investment, acting without evidence turns every budget commitment into a leap of faith.

So, how can we find facts before committing resources? It is not simply a matter of finding data. Nowadays, finding data that supports an idea is not difficult. What makes the difference between useful evidence and unreliable evidence is the method used to collect and interpret it. Although it may sound daunting at first, applying the scientific method in in business and product discovery simply consists of conducting experiments to test a hypothesis.

The real value of the scientific method is control. In business context, controlling an experiment does not mean to reproduce the rigor of laboratory science. It simply means respecting 3 principles:

1. Define the test process in advance. The process specifies what to test, how to test it, and what result would confirm or falsify the hypothesis.
2. Reduce bias and variance in the process. The test should produce reliable, objective evidence rather than simply confirm what we already believe. The test should also reduce unnecessary differences in how people interpret the evidence or make decisions from the same result. Unwanted variability in human judgment is what Daniel Kahneman describes as noise.
3. Considering alternative explanations. It is relevant to ensure that the observed evidence comes from the assumptions that are under test, rather than from other factors that are not visible.

But how to actually prepare and conduct rigorous experiments in a business context. And, more importantly, how to turn results into a decision?

This post brings together ideas from several established disciplines. The scientific method provides the fundamental logic: formulate hypotheses, design tests, collect evidence, consider alternative explanations, and update conclusions. The Lean Startup method applies experimentation to business models and product assumptions. Strategyzer provides practical tools to structure assumptions and test business ideas.

What this post adds is not a new methodology, but a practical synthesis focused on decision-making. It connects the different pieces into a step-by-step process to identify the assumptions behind a business idea, conduct experiments, and translate the results into a decision.

The method

Testing business hypotheses requires 6 steps.

1) Estimate impact

2) Identify assumptions

3) Rank assumptions

4) Define test approach

5) Gather data

6) Assess findings

1. Estimate impact

Business ideas can come from many sources. Some emerge from customer requests, others from competitors launching new products or features. Some come from observing macro trends, while others arise from a simple eureka moment.

Regardless of where the business idea comes from, it is important to define what the actual impact will be. For a startup founder this is a difficult step to do, because it requires to transform a business intuition into a quantified impact. Sometimes the impact may disappoints because. The point here is not to value an idea in absolute terms but in comparison with other ideas or other possible investments that a company or an entrepreneur could do. The simple logic is to concentrate effort in what appears the most value-generating idea.

Concretely, this step consists of estimating potential future cash flows that the idea will generate from day 1 until a certain day in the future. Cash flows are not only positive, such as revenue, but also negative, such as investment and running costs. Moreover, the time horizon of the impact is crucial because no idea can generate money forever. After all, on average a company lasts only 10 years before it is bought, merged, or liquidated, so there is a finite time any business idea can provide value for.

Moreover, impact is not always monetary, a new business idea could generate other non monetary benefit such as brand recognition. Nevertheless, it is important to quantify any qualitative benefit so that different ideas become comparable on monetary terms.

2. Identify assumptions

Once you formulate the business idea or hypothesis, before going out and test it, it is important to decompose it in the assumptions that may invalidate the business.

Imagine, for example, your idea is that “there is a new market for college education crowdfunding”. So you want to build a crowdfunding platform specialized in supporting students to find people that are willing to lend money to pay their tuition fees and be paid back after a few years.

Let’s suppose you have estimated the number of students in your country who need to fund their education. You have defined a business model based on a fee for each loan, calculated the costs to build and maintain the platform, and projected revenues and costs over the next few years. You now have an estimate of the business opportunity.

Now it is time to make it concrete and identify the critical assumptions behind your hypothesis. Here just a few of them, as an example:

• There are enough students who cannot afford to pay tuition fees for their desired education.
• Many high-potential students (with a low risk of default) are not eligible for, or fail to win, scholarships.
• There are enough lenders who are willing to invest part of their capital in exchange for a lower return than traditional banks, motivated by the prestige of funding education.
• Current crowdfunding solutions and charitable organizations do not provide a suitable solution.
• Students are willing and able to repay the loans after graduation.
• The default rate is low enough for the business to be profitable.
• The platform can legally facilitate student lending and comply with regulations (consumer-protection, know-your-customer, anti-money-laundering, privacy) at an economically viable cost.
• Students will trust an unfamiliar platform enough to disclose financial and educational information and enter into a multi-year financial obligation.
• The platform can acquire students and lenders at a cost that is materially below the revenue/gross margin generated per loan.
• The platform can reach sufficient marketplace liquidity: students can find funding and lenders can find suitable borrowers within an acceptable time.

As we can see from the example, assumptions tend to belong to four categories:

• Need — Is there a sufficiently important problem? Whether the underlying customer need or problem actually exists and is sufficiently significant to generate demand.
• Market — Is there an attractive market opportunity? Whether there sufficient customers and suppliers, and is the market not already adequately served by existing or alternative solutions?
• Solution – Does our proposed solution work? Whether the proposed solution can effectively address the identified need and Can the proposed product effectively solve the problem, and will customers adopt and use it.
• Financial viability – Can we make money sustainably? Whether the business can generate sufficient revenue to cover acquisition, operations, regulatory, and other costs while maintaining acceptable margins.

3. Rank assumptions

Before building the full platform, it is important to verify that the most important business assumptions hold. Usually, the most important ones are demand-based assumptions. In other words, if there is no market, there is no business.

Therefore, at this stage it is important to prioritize the assumptions based on the impact if they are wrong. Hence, for each assumption, the question is: “How big is the loss if this assumption turns out to be false?”

There are generally four levels of priority:

Critical – If this assumption is false, the business model is fundamentally not viable and there is no adjustment or workaround possible.

High – If this assumption is false, the business can still exist, but the expected business model or business strategy would require a major change.

Medium – If this assumption is false, the business would face meaningful challenges or additional costs. But these could potentially be addressed through operational adjustments.

Low – If this assumption is false, the impact on the overall business would be limited and could be addressed relatively easily without changing the core business model.

For the college education crowdfunding example, the most critical assumptions are five:

1. There are enough lenders who are willing to invest part of their capital in exchange for a lower return than traditional banks, motivated by the prestige of funding education.
2. There are enough students who cannot afford to pay tuition fees for their desired education.
3. Students are willing and able to repay the loans after graduation.
4. The default rate is low enough for the business to be profitable.
5. Current crowdfunding solutions and charitable organizations do not provide a suitable solution.

Now that the assumptions are prioritized, before going out and testing the most critical ones, they must be testable. Testability is what we define in the next step.

4. Define test approach

To be able to test an assumption reliably, it needs to be formulated in a way that is quantifiable and clearly falsifiable. In other words, it must be possible to determine objectively whether the assumption passes or fails the test.

Hence, for each critical assumption, we define a specific metric and a threshold that separates a positive result from a negative one. It is important to establish the test criterion before conducting the test, so that the result does not influence how the threshold is defined. For example, let’s say we want to test whether “enough students are interested.”

To make the assumption quantifiable and falsifiable, we can reformulate it as follows: “At least 20% of the target students in the surveyed sample would be willing to disclose their financial and educational data on the platform to apply for a loan.”

In the example, the 20% threshold creates a clear numerical boundary between a validated and a falsified assumption. The threshold allows to compare the test result with a predefined criterion and reach an objective conclusion. On the other hand, choosing the right threshold requires judgment. It should reflect the conditions that make the product or business model viable. For example, if showing the value proposition to new customers is expensive, the business needs a higher conversion rate to achieve a viable customer acquisition cost. Hence the threshold must be higher. Vice-versa, if getting the product in front of customers is cheaper, the business can afford a lower threshold.

Defining test criteria also requires to select a suitable testing method. Typically the testing method should be a good trade-off between rigor and easy of implementation (low-cost). Notable testing methods are customer interviews, customer surveys, landing page smoke tests, pre-sales, A/B tests, field observations, and so on. There is no general rule on what method to use. But each methods comes with its own level of reliability. In general tests that are based on observation rather than opinion are more reliable.

Finally, the other important aspect to consider when defining test criteria is sample size. It is the minimum amount of evidence needed to make the decision with an acceptable level of uncertainty and cost According to the Experiment Board, developed by Trevor Owens and Grace Ng from Javelin, a good practice is to define the minimum number of customers needed to gather meaningful evidence before running the test. A larger sample generally provides more reliable evidence. However, interviewing or surveying each additional customer requires time and resources. Beyond a certain point, each additional customer provides less new information. The goal is therefore not to reach the largest possible sample, but to define a minimum sample size that provides sufficient evidence at a reasonable cost.

5. Gather data

Once you define the assumptions and the test criteria, it is time to “get out of the building”. Test the assumptions one by one by getting customer data. Start with the highest-priority assumptions and go one by one.

Each test should produce objective evidence. Compare the results with the predefined threshold. Record the outcome as supported, not supported, orinconclusive. Also record the evidence that supports the result.

When a test falsifies an assumption, reassess the business hypothesis. Do not simply move to the next assumption. Finally, assess not only the data but also the process that led to the conclusion.

6. Assess findings

Once you have tested all the assumptions and found no critical falsified assumptions, reassess the business hypothesis based on the evidence you collected. The testing process creates a structured way to talk to customers and refine the idea. It helps you base the business hypothesis on realistic information rather than idealistic assumptions.

Something that often goes neglected is to evaluate the data as effectively explaining cause-and-effect without any other variable getting in the way. This is something to ensure during the data gathering.

For example, suppose students leave their contact details on the landing page used to smoke-test the crowdfunding platform. This result may suggest that they are interested in the value proposition. But how do we know that no other hidden variables explain their interest?

This matters because completing the actual loan application may involve a different process or a different price than the one presented on the landing page. These differences could affect the student’s willingness to proceed. We therefore need to check whether factors outside the test could compromise the result.

One way to do this is to interview a subsample of the students who expressed interest. Their answers can reveal whether other factors influence their decision and whether the landing-page test accurately reflects the solution’s marketability.

Choose the right method

The scientific method offers a rigorous way to test a business idea, but it does not always make sense to use it. The right approach depends on the importance of the decision and the cost of being wrong.

When the stakes are high, structured experiments can provide the evidence needed to make a sound decision. For lower-stakes decisions, such as choosing between two UX designs or testing the market impact of a small price discount, heuristics work better. In product discovery, heuristics are efficient mental shortcuts. They save time and effort by using only the information needed to make a good-enough decision.

Finally, if you have a product idea but are not yet ready to invest in it, feel free to reach out. I help teams choose the right approach to turn uncertain ideas into clear product and investment decisions.

Before you build, test what needs to be true for your business idea to succeed.

Photo by Cats Coming.