You don't necessarily need a massive sample size to make a solid conclusion, depending on the effect size. You can use statistics to control your error rates.