How would you A/B test sign-up button copy and measure results?

This tests whether you can design a rigorous frontend experiment beyond simply changing text. A strong answer covers four key parts: random assignment via stable bucketing such as user ID hashing; serving variant copy at render time without layout shifts; instrumenting both impressions and the sign-up conversion event; and analyzing for statistical significance rather than raw counts. Red flags include sequential testing or vanity metrics like clicks without measuring actual conversion.