To gain statistically significant results from a test, you need the right type of test and the appropriate level of volume. Multivariate testing procedures may require higher traffic than far simpler A/B tests. It’s generally preferable to obtain thousands of exposures to get accurate results with each test creative.