All major insurance companies share a database of all claims ever made going back many decades. When we do fire investigations, we document many things. The people who make the rules at insurance companies sometimes base their decisions on fallacious (cum hoc, ergo propter hoc) logic. They fail to recognize that correlation does not equal causation.
Let’s say, for example, an unlicensed incompetent person is doing illegal electrical work and having the homeowners get the permits. If he did 100 jobs and used a Crouse-Hinds panel on every job and later there was an electrical fire at every house, the fact that the houses each had a Crouse-Hinds panel would be documented. That may be the only discernible thing common to all the houses because there was no record of who actually did the work.
Some kid with a degree in accounting at an insurance company may be asked analyze the 100 claims. He sees that they all had Crouse-Hinds panels and guess what? Crouse-Hinds panels make it on to their Bad Stuff list.
Another in sitting in an office at another insurance company may look at 100 similar claims and recognize that there is no evidence pointing to the Crouse-Hinds panels as being the cause. Maybe he will conclude that because in every case the homeowner applied for the permit, that they shouldn’t insure homes where the homeowner presumably did the work.
In an effort to prevent making those bad decisions, insurance companies will randomly do in depth investigations where they wouldn’t normally do an investigation. I’ve done a few investigations where the loss was substantially less than the cost of the investigation but the insurance companies commissioned the investigations for statistical reasons.
From what I’ve heard from the field agents, the kids behind the computer doing the analyzing don’t often, if ever, communicate directly with the people in charge of the actual investigations.
Sometimes they get it right and sometimes they don’t.