If you know for a day of a test the data would be skewed somehow, for example by running a big marketing campaign, then it would be best to void the results for that day. When you void a test result, the result is no longer included in the overall test results. This helps to increase the reliability and accuracy of Splitly A/B tests.
From time to time we might experience problems our side of things or through MWS. If we do then we will void results for that day automatically. This is a very rare occasion and is usually nothing to worry about.