I understand that IQ tests give you a number but I have never really understood what is going on under the hood. Like what are they actually measuring, how do they turn a bunch of puzzle answers into a single score, and how do they know that score means anything? I also want to know how the norming process works because I keep hearing that scores are relative to the population but I do not fully understand what that means in practice.
What is actually happening under the hood is genuinely fascinating once you understand it 
IQ tests work by sampling performance across multiple distinct cognitive tasks that are known to load on general intelligence. The variety matters because g is a latent factor that you cannot observe directly, you can only infer it from the pattern of correlations across many different tasks. When someone does well on vocabulary, matrix reasoning, working memory, and processing speed simultaneously, the shared variance across those tasks is what gets captured as g. The raw scores from each subtest get converted to scaled scores using norm tables derived from a large representative population sample, and then those scaled scores get combined into composite indices and a full scale score. The norming process sets the population mean at 100 and the standard deviation at 15, so your score tells you where you sit relative to everyone else in the norming sample. A well designed assessment like RIOT IQ covers 15 subtests across 6 CHC domains which gives you both the overall g estimate and a detailed breakdown of your specific cognitive profile 
The norming part is what separates a real test from an online quiz 
Without a large representative norming sample your score is just a raw number with no reference point. Saying you got 47 out of 60 means nothing without knowing how the rest of the population performed on the same tasks.