Yes, the effect is usually stable for the same person on the same drug, so it's standard procedure to try a lot of antidepressants until one helps. Who claims that it's not? I've never heard of that, and it just sounds completely weird that a doctor wouldn't adjust treatment depending on results.

That's why averages aren't enough data. High-variance meds are better as long as the worst case is rarely lethal, since bad effects are felt only for as long as it takes to notice and switch whereas good ones are forever.