How to evaluate a business idea before you spend money

Describe a specific customer and problem

Start with a person in a situation, not a list of features. Who encounters the problem, when does it happen, and what do they do about it now? If your answer is “everyone,” choose one group you can learn from first.

For an illustrative appointment-booking service, “small businesses need software” is broad. “Independent repair-shop owners miss booking calls while working with customers” gives you something more concrete to investigate. It is still a hypothesis, not an established fact.

Write down the observation that led to the idea. Separate something you have personally seen from something you assume happens elsewhere.

Understand the alternatives, including doing nothing

People may already use a spreadsheet, a messaging app, an employee, or a competing service. They might simply tolerate the problem. Compare your idea with those real alternatives, not with an imaginary situation where the customer has no options.

• What is working well enough about the current approach?
• What is frustrating, slow, expensive, or unreliable?
• What would make switching worth the effort?

Research can help you discover alternatives and frame questions. A competitor's website cannot tell you why a particular customer would choose you.

Find the assumption that matters most

List what would need to be true for the idea to work. Then identify an important assumption you have little evidence for. Testing that assumption is usually more informative than polishing an unrelated feature.

Keep a simple distinction in your notes: observed, reported by a potential customer, estimated, or not yet checked. A confident sentence does not become evidence because you or an AI wrote it.

Choose a small test that can change your mind

Match the test to the uncertainty. To understand a workflow, ask people to walk through a recent example. To investigate whether an offer is understandable, show a simple description and ask what they think it provides. To learn about delivery, try a limited manual pilot with willing participants.

Before you begin, write what you expect to learn and what result would make you change direction. Avoid deciding afterward that every response confirms the idea.

• Ask about actual behavior and specific experiences, not only whether people like the idea.
• Record objections and cases that do not fit your hypothesis.
• Treat a small sample as a clue, not proof that a whole market behaves the same way.

Check the practical constraints

Consider how you would reach customers, deliver the offer, cover the work involved, and obtain any missing skills or resources. Record rough costs and unknowns separately; an unresearched number is an estimate, not a forecast.

For the booking-service example, a manual pilot may be manageable while a fully integrated product is not. That difference can change the first version you choose to test. Identify requirements that need qualified advice for your industry and location rather than treating a generic checklist as clearance to launch.

Choose a next move, not a verdict forever

Review what changed after the research or test. Continue if the evidence supports a useful next step; revise the customer, offer, or delivery approach if it does not. Pausing is also a legitimate outcome.

In 3Haystack, use Exploration to keep the outline, questions, and saved research together. Your personal excitement, confidence, and ease-of-launch ratings are a separate reflection—not a probability of success.

When the next step is clear enough to act on, use Launch to organize the work. The purpose of evaluation is to improve a decision, not to manufacture a reassuring score.