I am an instructor in PPCT (Pressure Point Control Tactics), of which pressure points are a big component. One of the things that I teach is that nothing in law enforcement is 100% effective 100% of the time. Second, there are certain points that are more effective in "pain compliance" when the person is sober and not actively fighting you (this is where most people complain the "system doesn't work" because they are using touch pressure to control a higher level of force it wasn't designed for. Lastly, in regards to your question specifically. Think of pressure points on a "bell curve"
Pressure points will work fairly reliably for about 70% of the population. About 15% of the population, you are going to get a REALLY significant response to them (we have all met the person in class who is barely struck and they complain on how hard you are hitting them). Then, you will have the 15% of the population that will have little to no response to them. This would include people who are high on adrenaline, drunk/high on a controlled substance, or just have a higher pain tolerance and don't respond as much. So, about 85% of people you would use pressure/pain points on, you will get a good response. But, if we think about most of the people we are going to be dealing with in a self-defense situation, we are more likely to be dealing with that other 15% who might not respond.
I hope that answers it without sounding like I'm skirting the issue, but it's the best I have to explain it and try to put some numbers on it.
The other thing I would point out is that some people teach "dim mak" solely based on acupuncture points and meridian theory. Others use it and also include the other areas that the Chinese taught. You have nerve points, muscle points, organ shots, bone shots, etc. Basically, they were attacking all of the body's weak points and it was all inclusive.