"Honestly do not understand how they [Google] make money."
By lying. If they collect thousands of datapoints, which they can prove, and hire a ton of machine learning experts, which they can also prove, then they must be able to use that to send ads to those who will most benefit from them, right? In the meantime, they just use the same crap algorithms based on browsing history and search term if applicable. Who knows what all the collected data is for, but eventually the guy who's responsible for thinking up the evil plan will come out and they'll do that.
This works for three reasons:
1. Google runs the ad system as a black box, so it is difficult for someone who pays for advertising to figure out who is really seeing the ads.
2. Companies are really bad at figuring out how useful their advertising budget is. Here's a good two-part summary of people who tried doing the research and all the problems they found, both in advertising itself and in advertisers' approach. It's a podcast but the pages contain transcripts for those who prefer to read text: Part 1 (mostly television advertising) Part 2 (online advertising)
3. Google has purchased almost all of their competition, and the others are either basically the same (Facebook) or didn't claim to be that smart in the first place (Bing ads). So you can't try out other advertising platforms to see if they can track better or don't need to, because you have no choices. Google's is biggest, so they get a lot of business.