Sometimes the world is a bit binary: if its not a 'genuine' mistake... then you must have done it on purpose! When journalists interview insurance companies you can almost hear the subtext out load: this wasn't a mistake, this is how you treat all your clients.
Sometimes it is bias, not even conscious, but an assumption about how the world works that means you might see some claimants as more likely to be fraudulent, in declining this one, that fits the pattern of the others, without examining all the evidence, means we make a mistake. It is simply not the same as the genuinely one-off mistake that can happen when a specialists report was overlooked.
Another is a complete mismatch in expectations. An insurer asks for a series of tests and reviews, which are "obviously" entirely standard in nature, to prove a TPD claim. The client sees an insurmountable list which is "obviously" intended to signal that their claim is unlikely to be paid and to deter them from pursuing it further.
Lastly, the sin the corporate is most likely to commit, is failure to take responsibility for the error - eloquently put in this post by Seth Godin.
That is why the category of mistake is really important to identify in order to build better systems and process - and to build more trust with clients and business partners.