The most likely cause of this error is that your sample size is too small. Only predefined sample sizes that are large enough create significant differences and meaningful results. If your sample (the number of visitors tested) is too small, the test loses its value.
We recommend you calculate the sample size you need to attain using a sample size calculator before running a statistical test. It is also recommended to wait a few extra days before consulting your results, even if your sample size is large enough, so you can account for user behavior on different days of the week.