Maybe it's a function of zip code, sex, age, profession, hobby, weight, etc... while another company will add a question about exercise level and put different multipliers on each of those.
They are constantly refining their statistical models to gain an pricing advantage over the competition without taking on more undo risk.
This applies to all kinds of insurance.
Combating fraud is ALSO a big thing. So they often share lists of customers they believe have defrauded their policies in the past, and are starting to use AI to find which patterns reduce the risk most, as well as getting around certain metrics they aren't allowed to ask about - like ethnicity.