Days 1 to 3: hunt for gripes, not compliments
Go where your target users already complain: Reddit threads, niche Slack and Discord groups, G2 reviews of incumbent tools, and support forums. You are looking for the same complaint phrased in different words by different people, ideally with evidence they tried to solve it. Save exact quotes, because the language people use to complain becomes your landing page copy. Tools like Olune automate this research pass by scanning communities and competitors for you, but manual reading works too if you put in the hours.
Days 4 to 8: run 5 to 10 Mom Test conversations
Talk to people who have the problem, and ask about their past behavior, not your idea. Good questions: when did this last happen, what did it cost you, what did you try, what did you pay for. Never ask 'would you use this', because polite people say yes and it means nothing. If 5 out of 10 people describe the same painful workflow unprompted, you have signal; if you have to explain why the problem matters, you do not.
Days 9 to 14: smoke test with real pricing
Put up a one-page site that states the problem, the fix, and a price, then send it to the communities and people from week one. Showing pricing is the point: an email signup for a vague free thing tells you almost nothing. A reasonable benchmark is 20 to 30 percent of visitors from a warm audience clicking a priced call to action. Spending $100 to $200 on ads can top up traffic if your communities are small.
The strongest signal: money before code
Ask for a pre-order, a 50 percent deposit, or a signed pilot agreement before you build anything. Three strangers paying $20 beats three hundred email signups. If nobody will pay for a discounted early version, they will not pay for the finished one either. That result stings for a day and saves you six months.