I have launched more than ten products from zero and watched many more ideas from the inside of large companies. The pattern I trust most is boring: the teams that win talk to customers before they write code, and keep talking after.
The expensive way to learn
Code feels like progress. It compiles, it demos, it gets screenshots on LinkedIn. But code is the most expensive way to discover that a problem is not painful enough. Every week spent building on an untested assumption is a week of runway spent proving yourself wrong slowly.
Twenty structured conversations with the right buyers will usually tell you three things code never will: whether the problem is real, who owns the budget, and what they do today instead.
What a good validation conversation sounds like
- Ask about the past, not the future. "Tell me about the last time this happened" beats "Would you use a tool that…?" People are generous with hypotheticals and honest about history.
- Find the workaround. If they have built a spreadsheet, a script or a manual process to cope, the pain is real. If they shrug, it isn't.
- Follow the money. Ask who would approve spending on this and from which budget. A problem without a budget owner is a hobby.
- Listen for urgency words. "Every week", "audit", "we lost a customer" signal priority. "It would be nice" signals a feature, not a company.
Turn conversations into evidence
After each call I write down the problem in the customer's own words, the current workaround, the cost of that workaround, and who decides. After ten calls the clusters become obvious. After twenty you can usually write a one-page problem statement that a stranger in that role would nod along to.
Only then do I sketch a solution, and the first version is rarely software. It is a clickable prototype, a concierge service or a spreadsheet I run by hand for three customers. If they come back for more, I have earned the right to build.
A simple scorecard
Before committing engineering time, I want honest yes answers to four questions: Is the problem frequent? Is it expensive? Is there a clear buyer with a budget? Do at least a few people want to try a rough version now? Two out of four is a research project. Four out of four is a product.
Fall in love with the customer's problem, not with your first solution. The solution will change three times; the problem is what you're really building around.
Startups are a race to learn faster than your runway burns. Conversations are the cheapest learning there is. Book twenty before you write the first thousand lines.
