Startup needed to launch MVP fast. Engineering wanted to build A/B testing infrastructure first. I pushed back.
THE DECISION
Launched without A/B testing, added it 3 months post-launch.
Building proper A/B testing framework would delay launch by 4 weeks
MVP was about validation, not optimization
Could make decisions based on qualitative feedback initially
Only 2% of startups fail from lack of A/B testing; many fail from launching too late
Launched 4 weeks earlier
Simpler codebase
Less analysis paralysis
Made some decisions on gut vs data
Harder to add A/B testing later
Missed some optimization opportunities
Launched, got users, validated core value prop. Added A/B testing when we had enough traffic to make it meaningful (3 months in). By then we knew what to test.
Infrastructure is important, but only after you've proven people want your product. Ship first, optimize second.