Use this calculator to figure out how big your test segments need to be. First, pick your confidence level. If you choose a lower confidence level, you can test a smaller volume, but there will be more possibility that any difference you see is due to chance.
Second, put in the response rate you anticipate for your test, and then enter the response rate you anticipate for your control. We will use that information to recommend a sample size necessary for a statistically significant result.