# 2 · Problem: what's wrong, and what better looks like A request usually arrives as a solution: "add a dashboard", "build an app that…", "we need notifications". Behind it is a problem someone noticed. If you build the solution without knowing the problem, you can build it perfectly and still miss. This part assumes you know who it's for (`1-who.md`). --- ### P1 · Find the problem behind the request **Rule:** For every product or feature request, write the problem it's meant to solve: what goes wrong today, for whom, and when. If the request doesn't say, look for it in what they said around it. If it's still missing, ask once: "What happens today when…?" or "What would this let you do that you can't now?" If you see a simpler fix for the same problem, offer it in one line with your recommendation, then build what they choose. **Why:** People describe the first fix that came to mind. "Export to CSV" may really mean "my accountant needs these numbers every month", and a monthly email to the accountant solves that with no spreadsheet at all. Knowing the problem also tells you what "done" means. **Agent trap:** both extremes. Building the stated feature literally and completely, without asking what it's for. Or deciding you know the real problem and quietly building something else. Both skip the one conversation that matters. **Check:** can you finish this sentence without naming a feature: "Today, struggles to … when …"? If you'd suggest a different fix, did you offer it rather than switch to it? > **In ihumane Pro** (https://ihumane.ai/#pricing): > - P2 · Say what success looks like, and how they'd notice > - P3 · Name the one idea that makes it different > - P4 · Check the problem is worth a product ## Check - [ ] The problem is written in one sentence, without naming a feature. - [ ] A simpler fix, if you saw one, was offered in one line, not silently swapped in. - [ ] Success is something observable, with a way they'd notice; no numbers they didn't give. - [ ] The one idea is stated, and a competitor couldn't use the sentence unchanged. - [ ] You know how often the problem happens, and considered whether a smaller answer would do.