So OP claims that data-driven ads are a scam using the anecdata that she is getting served irrelevant ads.
I'm not going to argue about whether ads are "worth it", but working on the other side (serving and optimizing ads), I can tell that yes, we run the numbers, and the math (mostly) works out.
As per your examples, we have run experiments by applying targeting ads based on narrow segments, and others without. As it turns out the data shows that "casting a wider net" works better, statistically.
I'm not going to argue about whether ads are "worth it", but working on the other side (serving and optimizing ads), I can tell that yes, we run the numbers, and the math (mostly) works out.
As per your examples, we have run experiments by applying targeting ads based on narrow segments, and others without. As it turns out the data shows that "casting a wider net" works better, statistically.
Make of it what you will.