How Do You Decide How Many People to Send to Which Experiment?

We will often set up a b tests, meaning we send 50 % on the users to one experiment. We send the other 50% to another where only no one thing is slightly different. And then it's two experimental groups. So when you say you've introduced two things, does that mean you've made one group see a change, and another group see the same old thing? I mean, it just depends what it is. It doesn't go about it necessarily in a scientifica lay o.