A business idea is a guess about a problem, a customer, and a way to help. Your first job is to turn that guess into evidence. You can do that with a few evenings, a notebook, and conversations. You do not need a logo, a big inventory order, or a finished website yet.
Start with a customer, not a category
“Small businesses” is too broad to guide a first offer. “Independent dog groomers who lose bookings when they cannot answer the phone” gives you somewhere to start. Name who experiences the problem, when it happens, and what it costs them in time, money, or frustration.
Write one sentence: I help [specific customer] solve [specific problem] by [small, concrete service]. Treat it as a working hypothesis. You are allowed to change it.
Have ten conversations about the past
Ask about the last time the problem happened. What did the person try? Who handled it? What did they pay? A recent example tells you more than “Would you buy this?” People can be encouraging without intending to spend.
Keep notes in the customer's own words. Separate observations from your interpretations. “Three people hired a temporary assistant last month” is evidence. “Everyone wants automation” is an interpretation.
Ten conversations are a useful first batch, not a statistically valid market survey. If every person is a close friend, broaden the group. Look for people with the same problem who do not need to protect your feelings.
Look at the alternatives
Your competitor might be a spreadsheet, an employee, a family member, or doing nothing. Find out why the current approach is acceptable and where it fails. If the problem is mildly annoying but never urgent, it may be difficult to sell a solution.
The SBA's business planning resources are a useful starting point for organizing market research. Add your direct customer evidence rather than relying only on an industry headline.
Offer a small, paid pilot
A pilot should have one deliverable, a clear price, a realistic deadline, and written expectations. For example: organize one month's receipts for a fixed fee, rather than promising to transform someone's entire operation.
Be honest that this is a pilot. Explain what you can deliver, what is excluded, and how cancellation or refunds work. Do not collect payment for something you cannot reasonably fulfill. For regulated work, check licensing and insurance before offering it.
A deposit or a completed purchase is stronger evidence than a waitlist signup. Even a purchase is only one data point: record whether delivery was profitable and whether the customer would buy again.
Set the decision rule before you see the result
For an illustrative first test, you might decide to continue only if several people describe a costly recurring problem and at least one pays for a narrowly scoped pilot. This is your test rule, not a universal success threshold.
If people care about a different problem, change the offer. If the offer takes twice the time you expected, change the scope or price. If the problem is weak, stop before spending more. Stopping a poor idea is useful progress.
Your next step
Choose one customer group and schedule your first three conversations. Use the validation workbook to record the problem, existing alternatives, and your next test. Keep your job and your options while you learn.