It can make sense to trust an account based on its NIP-05 identity if you know the domain.
But the checkmark is shown besides the name of just any account that is associated with any domain name, trustworthy or not. Any scammer or spammer can trivially create a domain name and they are often very willing to pay. So the mere presence of a NIP-05 identity proves nothing.
A checkmark doesn't signal that the user needs to manually verify that the domain is one they trust. The message it conveys is that the account is to be trusted. And, again, the presence of a random domain is no evidence for this.
A solution could be to let users specify a pool of domain names that are to be trusted and have the checkmark for those domains.
The issue is that NIP-05 only allow each user to have one identity. So if there are two domains that could verify me, for instance because I belong to two organizations (many people do), I have to pick one and only one. The people who don't know the organization that I picked, but do trust the other, won't see a checkmark, even though they logically should.
Even for its intended purpose, NIP-05 should allow multiple identifiers. But verification is not its intended purpose.