"We can't afford research" usually means "we can't afford a six-figure lab study." Fair. But five conversations cost almost nothing — and they catch the disasters. This is the exact playbook I run before designing anything.
Recruit in 48 hours
Screen for behavior, not demographics: "bought shoes online in the last month" beats any persona slide. Offer a modest incentive, over-recruit by two, and run sessions remotely with screen share. Your own customers, a rival's reviewers, or a panel — in that order of preference.
Test tasks, not opinions
Never ask "would you use this?" Ask users to complete three real tasks on your current site or a prototype, and watch where they hesitate. Hesitation is the data. Five users uncover the vast majority of critical issues — the sixth mostly repeats the fifth.
Report in one page
Severity-rated findings, each with a clip, a quote and a fix. Founders read one page; nobody reads thirty. Ship the fixes, re-test the top two issues, and you've done more research than most funded startups.
Total cost: two weeks and a few incentives. Total value: knowing you're building the right thing before the expensive part starts.