04. User Research

Chapter 4 of 24 · 15 min

KEY INSIGHT

The best user research method in this market is watching someone use your product (or try to). Conversations are useful, but behavior reveals truth that interviews hide. Before building anything, conduct problem validation interviews. Recruit 5-10 people who match your target user. Pay them ₦5,000-₦10,000 for 30 minutes. Don't pitch your solution—ask about their current workflow and problems. Interview structure that works: 1. Context setting (5 min): "Walk me through how you currently handle [relevant task]" 2. Pain exploration (10 min): "What's the most frustrating part of that process?" 3. Current solutions (5 min): "What have you tried before? Why did it work or fail?" 4. Validation check (5 min): "If a tool existed that [specific capability], how would you use it?" 5. Payment reality (5 min): "How much would you pay monthly for a solution that solved this?" The payment reality question is the most important and most avoided. Users consistently overstate willingness to pay. Watch what they do, not what they say. ```python def research_synthesis(interview_notes): """ Structured analysis of qualitative research data. Extracts patterns from multiple user interviews. """ # Categorize pain points by frequency across interviews pain_frequency = {} for note in interview_notes: for pain in note['pains_mentioned']: pain_frequency[pain] = pain_frequency.get(pain, 0) + 1 # Extract willingness to pay data wtp_values = [note['stated_wtp'] for note in interview_notes] avg_stated_wtp = sum(wtp_values) / len(wtp_values) # Adjust for overestimation (real-world adjustment factor) adjusted_wtp = avg_stated_wtp * 0.6 # Nigerian market factor # Identify blockers for adoption common_blockers = [] for note in interview_notes: common_blockers.extend(note['blockers']) blocker_count = {} for blocker in common_blockers: blocker_count[blocker] = blocker_count.get(blocker, 0) + 1 return { 'top_pains': sorted(pain_frequency.items(), key=lambda x: -x[1])[:5], 'stated_avg_wtp': avg_stated_wtp, 'adjusted_wtp': adjusted_wtp, 'top_blockers': sorted(blocker_count.items(), key=lambda x: -x[1])[:3], 'sample_size': len(interview_notes) } # Example synthesis sample_notes = [ {'pains_mentioned': ['time', 'cost', 'accuracy'], 'stated_wtp': 5000, 'blockers': ['trust', 'payment']}, {'pains_mentioned': ['time', 'accuracy'], 'stated_wtp': 8000, 'blockers': ['learning_curve', 'payment']}, {'pains_mentioned': ['cost', 'accuracy'], 'stated_wtp': 3000, 'blockers': ['payment']}, ] synthesis = research_synthesis(sample_notes) print(f"Adjusted WTP (accounting for overstatement): ₦{synthesis['adjusted_wtp']:,.0f}") print(f"Top blocker: {synthesis['top_blockers'][0][0]}") ``` Remote research has specific Nigerian failure modes: poor video call quality, sudden disconnections, respondents multitasking. Build in redundancy—record sessions, take notes, follow up via WhatsApp voice notes for clarification.

User research for local AI products requires methods adapted to Nigerian communication patterns and infrastructure constraints. What works in Silicon Valley focus groups fails here—users are busier, less familiar with "sharing feedback," and more practical about their time.

EXERCISE

Conduct three problem validation interviews using the structure above. Document insights in a shared document. Identify the most common pain point and whether users can articulate willingness to pay.