Meanwhile, my insurance company would have done what, exactly? Would they have even asked me why I was speeding?
I don't think you get what a sophisticated data model is capable of. If safer drivers do really need to go 30% over the speed limit occasionally like you suggest, the data should show it. People who never go fast, would then get punished, as would people who would go over the speed limit very often because they are reckless drivers.
Given a large enough sample set, the insurance companies don't need to know the particulars of why you were speeding in a particular way once, it is expected that good drivers will, they just need to know if you are doing it more often, or in a particular way, that they can say with statistical significance that you are more likely to have an accident. The occasional incident that you suggest is not going to allow them to say anything with any amount of statistical significance.