The author should plug the numbers in here: http://www.usereffect.com/split-test-calculator
I'd go as far as to consider the data to be worthless without the confidence value