so you could check the value of a feature (1) with an experiment and then if you're served variation A, then check the value for feature 2. You could do this pretty cleanly by setting the value from feature 1 as an attribute, then targeting the second experiment to those with the right value for that attribute.