CRO beyond A/B tests — what big teams skip
A/B testing is the obvious tool. Three less obvious moves that beat it on small samples.
A/B testing assumes you have enough traffic for statistical significance. For sites under 5000 visits per week, A/B is too slow — you'll wait three months for one decision.
What works on small samples:
- Heat maps and session recordings (Hotjar, Microsoft Clarity). 50 sessions tell you where people get stuck.
- 5-second tests on UsabilityHub. Shows page for 5 seconds, asks what they remember. Catches first-impression failures.
- Exit-intent surveys. 1-question popup before they leave: "What stopped you from continuing?" Real-time qualitative data.
Use A/B tests for high-confidence decisions on high-traffic pages. Use the diagnostic tools above for everything else.
The order matters: diagnose first, hypothesize, test. Skipping diagnosis means testing random ideas — slow and expensive.